AnywhereAmps on Schallbert's Blog

USB Power Delivery

Schallbert — Thu, 27 Jan 2022

USB Power Delivery

USB Power Delivery is a standard within the USB 3.x protocol that specifies power supply and voltage levels within the USB network. With USB-PD it is possible to daisy-chain consumers which then negotiate power demands with the source(s) automatically. The standard has different versions, of which the latest (2021) allows up to 5 Ampères at 3.3…48V and a maximum of 240 Watts of power.

USB-PD roles

In the USB Power Delivery Standard, there are different roles that participants on the ends of an USB cable can play. They can be host or client for data purposes, and at the same time power source, power sink, or both for other participants.

Role terminology

USB-C uses the acronym DFP or Down-Facing Port to make clear that the port is the one of a Host. UFP or Up-Facing Port is used for clients that can be connected to hosts. There’a third role, though: DRP or Dual-Role Port. This port can serve either as a host or as a client, depending on who the connecting partner on the other side of the cable is and what the USB network negotiates.

In USB-PD context, any of those ports can be configured to be sinking or sourcing current. Example: Although a Laptop’s USB-C port’s role may be configured to DFP, when connected to a power source, it can still be a power sink. Hmm.

In this post, I’m only using the UFP role in both USB-C and USB-PD word sense, meaning that I’m trying to configure a power sink for my device so that it will get the input voltage and current it needs.

USB-PD version

Before USB-PD Standard 2.0, USB-C devices only support so-called Power Profiles with fixed voltages of 5V, 12V, 20V and currents of up to 2A, 3A, or 5A.
Between USB-PD Standard 2.0 and USB-PD 3.0, Power Delivery Objects replaced the old Power Profiles that allow different voltages/currents to be selected by the sink depending on the power source’s offerings. Voltages are fixed to 5V, 9V, 15V, 20V with current limits of 2.25A, 3A, 5A depending on source capability and voltage setting. PDOs allow fallback strategies, so if a source cannot support the highest power demand, the sink may select another PDO with less voltage/current that works for both source and sink.
With USB-PD Standard 3.0 PPS , the protocol offered even more flexibility with Programmable Power Supply. Here, a power sink requests a voltage level between 3.3V and 21V, selectable in 20mV steps and 0.2 to 5A, selectable in 50mA steps while staying fully backwards-compatible to USB-PD 2.0’s Power Delivery Objects.
Latest USB-PD Standard 3.1 even enhances Programmable Power Supplies’ limits and allows 3.3V - 48V and up to 5A using AVS - Adaptable Voltage Files. Whow.

Not all sources (wall plugs, powerbanks etc.) can supply all Voltage levels or currents, or even share the same standard of implementation. But a number of values is pretty common: 5V, 9V, 12V, 15V, 20V. Some of the products I have seen on the market offer up to 4A, but most keep it with up to 3A and voltages up to 15V are most common (in 2021).

Power delivery ASICs

ASIC stands for application specific integrated circuit, and in this post I’m trying to shed some light on ASICs with Power Delivery sink role that can work stand-alone, that is without an external control device such as a microprocessor. I found three types of these devices:

Standalone USB-PD sink, voltage and current setting via resistor cascade
Standalone USB-PD sink, voltage and current map programmable via data bus (I2C/UART/SPI) into non-volatile memory of the ASIC
Standalone USB-PD sink with integrated microprocessor that needs to be programmed in order to work properly

What all ASICs have in common: They require peripheral circuitry to work - at least some resistors, capacitors, and power transistors to switch power supply and negotiate with connected power sources.

USB-PD is complex: Versioning and power contract negotiations

Both the downwards compatibility and the scaleability of USB-PD take their tribute - the USB-PD specification is a >600 page monster and it’s not always clear which source or sink supports which protocol level.

As both the source, sink, and the cable have to negotiate a Power Delivery contract based on the sink’s requirements, the source’s power options, and the cable’s capabilities, you can imagine that this process is very complex. Even more so as all participants may have implemented a different USB-PD standard, revision and version. Fun fact: The USB Power Delivery specification package’s zip folder on the official page has a file size of >42MB. Although I think this topic is very interesting, I don’t have the time to deep-dive into the specifics 😅

So, let’s leave all this stuff behind and try finding an easy-to-implement solution.

Example ASICs

I have arbitrarily chosen three ICs from the web, supplied by different companies, one each per device type:

Cypress/Infinion’s CYPD3177, a resistor-config USB-PD sink controller
STMicroelectronics’ STUSB4500, a Memory-config USB-PD sink controller
Microchip’s UPD301C, a fully programmable USB-PD dual role controller

I’ll select the STUSB4500 first.

Chip / IC Shortage

For integration into my design, I wanted to buy some samples. Unfortunately, Chip shortages hit hard so that neither of the distributors that I normally order electronics parts with, had any in stock for the next year.

That’s why I decided to order a development board with the ASIC readily mounted. This is 10x more expensive than the chip itself but I didn’t have a choice here.

Integration of STUSB4500

STUSB4500 is using the USB-PD Standard 2.0 - so not the latest greatest one - but downwards compatibility of later standards make it a quick and easy starting point for a maker. The device comes pre-programmed with three Power Delivery Objects:

PDO1 = 5V/1.5A
PDO2 = 15V/1.5A
PDO3 = 20V/1A

So for my AnywhereAmps Alpha 1.1 evolution 1.2 I can use this thing right out of the box without re-programming. If I wanted to change the Power Delivery Objects, I’d use an Arduino board with I2C and install STUSB4500 library which is readily available e.g. at Platformio. With this, I could even activate USB-PD 3.0’s Programmable Power Supply if I wanted. Let’s see, when I find the time, I’ll prepare a specific blog entry for programming STUSB4500.

Powering up the amplifier

OK, so I connected the eval board to a 15V capable 30V USB-C power supply. The purple light indicates 15V available on the output rail.

Now I attached my amplifier circuits at its output terminals for an initial test. STUSB evaluation board lights gone, amp off. Meh. At least I have an indicaton for what the cause might be: It’s capacitive inrush currents, again! To be able to properly power devices with large bulk capacitors, either I’ll have to set current in the PDO to abormally high levels or I make up my mind aboud constructing a proper, miniaturized inrush limiter circuit. I have setup a prototype already that I discuss here.

Should you use USB-PD for your projects?

The answer is a definete “it depends”. Although the capabilities of USB-PD are great, there are a few drawbacks:

requires an extra ASIC with periphery to negotiate the power contracts. This takes PCB space and increases project costs significantly
To be fully backwards compatible, it makes sense to buy an ASIC that both supports Standard 3.x, and ideally, you take one that is programmable
If you take a programmable device, there’s an additional learning curve for software development, the ASIC-specific libraries you want to use, and possibly even the need for programmer hardware
You have to make sure that power supply and cable are compatible to your USB-PD sink’s needs. Cables for high currents are expensive, too
USB-C port is miniaturized. SMD soldering and entry-level soldering stations might not become friends ever, so you should bring some experience on that topic.
Barrel-jacks are just round and thus much easier to mount in an enclosure than a PCB with attached USB-C socket that also has to withstand considerable physical strains.

On the other hand, using USB-PD makes much sense when you

have a mobile gadget that you want to take along with you, or that is often connected to and disconnected from power sources
want the flexibility of battery-powered devices: Just connect a compatible power bank and off you go
need a data connection to your device anyways
have special constraints on your power supply and want very specific voltage/currents to be delivered
want easy overcurrent and overvoltage protection out of the box

Metabo / Cordless Alliance Systems

Schallbert — Thu, 09 Dec 2021

Power Source for AnywhereAmps

Status quo

I have used Metabo / CAS (Cordless Alliance System) batteries for both AnywhereAmps Alpha 1.0 and Alpha 1.1. The latter already uses the power supervisor IC MAX17562, protecting both battery and circuits from reverse current, overcurrent, and undervoltage.

Problem

Although using the power supervisor IC was a big leap forward, still it would be better if I was able to read battery state of charge directly. This wouldn’t just enable safe operation, but also allow auto-switching the amplifier off when the battery goes low without side effects like auto-restart when battery voltage jumps back up when load current drops on shutdown.

Discussion with Metabo/CAS

Metabo logo ^[1]

That’s why I contacted the Metabo CAS team, asking whether I could get some support and the specifications for communicating with the battery pack. I was quite happy to see that they respond really quickly, and soon I was on a phone call discussing the matter.

Outcome:

Usage of CAS batteries in other devices than the alliance’s power tool families is considered a “hack” and thus wouldn’t be approved by Metabo
Metabo is interested in cordless tools manufacturers to join their battery alliance, but my application in an amplifier is just too far away (Metabo’s slogan once was “Work, don’t play”)
Metabo would expect me to produce 10k+ units per year before I could joind the alliance - which is just unrealistic
I’d have to understand, but sharing the specifications and allowing me to use the interfaces would draw some of their resources while they wouldn’t have any expectations on return of invest

As disappointing 😔 as this is, it didn’t come totally unexpected. The engineer on the phone made time for me. He told me that they’d appreciate my enthusiasm about their battery systems very much and that I shouldn’t feel discouraged.

Well. Let’s talk about an

Alternative

I know only one other power standard also featuring battery powered systems that meets my expectations below.

Requirements

“open enough” for my usage
safe
of high quality
meets my purpose’s voltage and current demands
widely available

USB-PD logo ^[2]

USB Power Delivery

The standard is called USB Power Delivery. Although it went through my mind earlier, I never wanted to actually use this because it involves massive additional work.

USB Power Delivery uses a complex protocol to establish a power contract between source and sink devices. All this has to be developed, a specialized IC has to be selected and peripheral circuits need to be designed for this to make it work with my setup.

😱 Oh no, they are out of stock!

I’ll write a separate blog post once I got a grip on this new subject.

Configure MAX1756x for your application

Schallbert — Sat, 04 Dec 2021

Power Supervisor configuration made simple

I use MAX17562 ICs in my design for AnywhereAmps Alpha 1.1. For you, I made a quick summary of MAX1756x features and how to configure them. 🙂

Of course, the datasheet is the master of all information I provide below. I’ll just add some design considerations, details on my way of using it, and provide application proposals.

What are they, again?

MAX1756x is an adjustable overvoltage/ undervoltage and overcurrent protector IC. It is available in three different versions for handling over current events:

MAX17561: Will turn off in an overcurrent event and autoretry after typically 600ms
MAX17562: Will latch off in an overcurrent event; restart after enable cycle
MAX17563: Will continue normal operation with current limiter active (dissipates heat)

Both versions are limiting overcurrent events to the configured current limit.

Use my MAX1756x application tool!

To simplify usage of Max1756x IC in your application, you may use my Max1756x application calculator spreadsheet.¹

MAX1756x current limiter / overload detector

Courtesy of Maxim Integrated: SETI resistor config ^[1]

The IC features an adjustable current limiter and overcurrent detection (SETI). When current increases above SETI threshold, it is limited to the set value. For MAX17561 and MAX17562: In case this condition remains true for more than typically 20.7ms blanking time, the device will turn off the load circuit. As the MAX17563 will continue normal operation, its power dissipation might lead to over temperature shutdown at some point.

This is how you configure the current limiter / overload detection using the SETI input: Select a resistor based on the curve on the right (taken from the datasheet) and connect it to SETI and GND. For AnywhereAmps Alpha 1.2, I’m using a 3k3 resistor to get a current limit of ~3.5A.

MAX1756x undervoltage configuration

The IC features undervoltage lockout (UVLO). When voltage decreases below UVLO threshold for typically 16.7ms, the device’s power transistors will switch off the load circuit. Internally, the IC will compare the input voltage at the UVLO pin with the reference voltage Vbg = 1.2V. To set the undervoltage lockout threshold, use a voltage divider to keep UVLO level above Vbg during normal operation.

MAX1756x overvoltage configuration

The IC features overvoltage lockout (OVLO). When voltage increases above OVLO threshold for typically 16.7ms, the device’s power transistors will switch off the load circuit. It works identical to the undervoltage lockout with the only difference that voltage at the OVLO pin should remain below Vbg under normal operating conditions.

MAX1756x reverse current protection

The IC’s reverse current protection can be enabled by pulling the *RIEN pin to GND. If you intend to use this feature, remember to not accidentally use polarized bypass capacitors like I did at the IC’s input as they will explode if you apply reverse voltage.

Fault *FLAG capability

MAX1756x will indicate a power fault using its *FLAG port under the following conditions:

Thermal overload
Overvoltage lockout defined through OVLO
Reverse current: when reverse current protection *RIEN has been enabled
Overload current defined throughSETI

Recommendation: Use TVS diodes

As the datasheet recommends and my experience from incidents in the past shows, I’d use 27V Transient Voltage Suppressor diodes in parallel to both MAX1756x’s input and output ports. It will protect (any inductive) load circuitry from negative voltage spikes on shutdown and the IC itself from surge voltages when connecting a live power supply. Plus, it will clamp reverse input voltage to non-harmful levels.

Recommendation 2: Use non-polarized input capacitors

To mitigate running reset issues when connected to “weak” power supplies or high-resistance input power wiring, use bigger non-polarized input bypass capacitors than you normally would (no, 100nF might not be enough!) - Especially if you drive capacitive or other loads that cause input surge currents.

For more information how to avoid problems here, please read my knowledge blog entry about inrush current limiters.

Disclaimer: Although I thorougly read the datasheet and carefully added the data, I can of course take no guarantee whatsoever for the results you yield from this tool. ↩︎

Power supervisor MAX1756x on capacitive load

Schallbert — Tue, 02 Nov 2021

Argh, I killed parts of my Power&Preamp circuit for AnywhereAmps Alpha today. During testing.

Incident

I accidently switched VBat and GND on my PCB. You shouldn’t think this happens at all, but it did again. The power input/bypass capacitor blew up instantly but my supervisor chip protected the rest of the circuits.

Repair and tests

After repairing with a ceramic capacitor, I continued some tests.

Reverse polarity test - PASS
Hard short to GND test - PASS
Undervoltage lockout - PASS
Normal voltage retry - …

Irregularity

But then, when I raised voltage slowly to escape the undervoltage lockout the system wouldn’t turn on, but still current consumption went up. To analyze, I connected an oscilloscope to see what was going on. You can see that the power amp’s >1400µF bulk capacitance acts like a soft short to GND so that the power supply voltage drops by 2V within 50µs to the undervoltage lockout trigger level. The supervisor IC then sheds the load, which makes the supply voltage jump back up. I suppose the ringing comes from the power supply’s regulator.

After the blanking time (datasheet MAX1756x) passes, it retries to switch on and the whole process starts over again.

Incident #2

When I measured voltage at the VCC rail supplying preamp, power amp, and 9V voltage regulator during unsuccessful switch-on, I noticed that negative voltage spikes occur at around -4V when the supervisor IC switches off, caused by the inductors in the power amp supply lines. While I was still wondering that these running resets might not be good, my preamp IC exploded. Hm.

What I learned

Never use a polarized capacitor as bulk/bypass “in front of” your protective circuits!
Use larger-than-100n bypass capacitors before voltage sensing devices when switched loads are capacitive. It helps stabilizing the voltage locally.
To protect the output side of the switch, add a Schottky or transient voltage suppressor (TVS) diode that clamps potentially harmful voltages
Limit in-rush current (follow-up article) as a precaution. I’ll switch to MAX17563 for the next circuit revision. It has a programmable current limit so that input voltage doesn’t drop too low in the first place.

Protect from negative voltage spikes

For my application, a fast-switching clamp diode in Schottky design is enough because behind the MAX1756x IC, the circuit will not see overvoltage spikes but reverse voltage spikes (due to inductors in switching regulator and amp design) only. In this article, I’ll explain the Transient Voltage Suppressors in detail.

My Printed Circuit Boards arrived!

Schallbert — Sat, 16 Oct 2021

Sequel of Printed Circuit Board design blog entry

Arrival

After less than two weeks, my PCBs finally arrived along with all the parts needed to solder. I think they look beautiful :)

Soldering: Step-by-step

I decided to solder the PCB in intervals, starting with the power management section. This way, I can test and correct the sections independently from each other. The SOIC (small outline integrated circuit) outline can still be hand-soldered easily when using lots of flux and a tool to “zoom in” for optical inspection. The only issue was that the IC has an “exposed pad”, i.e. there’s a ground pane on the bottom of the IC for better heat conduction. Soldering this was a nightmare. Finally I managed to fiddle solder through one of the bottom vias while keeping the iron at another via vor nearly a minute until it finally worked out fine. I’m pretty happy with the result, though. The system can be activated pulling *HVEN low, lighting up the LED. Tests for undervoltage/short circuit behavior went very well - the battery pack is protected now.

The Preamp section was easy to solder as well. Only the polarised capacitors proved to be so big that they covered much of the solder pad - this is a problem because the solder wouldn’t want to connect both cap and pad but just stick to one of these.

The 9V buck converter section - well… I didn’t manage to solder it by hand at all. Quadruple Flat Packs without leads (QFN) are hard enough, but this one has an exposed pad as well and I had to prepare much ground area to get the heat away from this component, so I couldn’t even get the temperatues high enough. So I borrowed a Hot Air station, and after some anxious trying and drowning the components in flux, I finally did it.

I used my digital camera with a macro objective to inspect the solder connections.

Working

I build a lowpass and a highpass into the design to support 2-way speaker systems. To test the three preamp outputs (fullrange, lowpass, highpass), I directly connected the preamp to my speaker. It is not a good idea to do this, because the impedance of the speaker is far lower than the one of the OP-amp, so the output power is really low. Plus, the lower frequencies consume most power so the sound you hear is “high-pass” already and the low-pass volume is considerably less. Still, it works and I can adjust the gain as intended.

Your browser does not support the video tag.

Preamp test

That’s it, my design is working! And remember, it is open source so you can download the project from my Github repository and order your own copy.

Lessons learned

Here’s the pitfalls and design flaws that should be corrected for Revision2:

Prefer to place all components in the same direction (e.g. horizontally)
Use thermal reliefs as much as possible to mitigate soldering issues due to thermal conduction
Exposed pads: When hand soldering, the according solder pads must be extended to the sides of the IC for better solder heat conduction
Avoid QFN footprints when hand soldering. They are a pain.
If possible, don’t use footprint with “exposed pad” or “power pad”. A hot air station might be mandatory.
Try to leave more space between component pads, esp. if placed horizontally.
As SMD caps have more height, allow for bigger solder pads.

FreeCAD modeling workflow

Schallbert — Tue, 12 Oct 2021

Motivation

FreeCad logo ^[1]

I just received a second order for AnywhereAmps Alpha. My initial prototype already has a new home, so I’ll be building at least three more copies. All the woodworks done by hand would take 4-5h for each copy, so I decided I want to limit the woodworking stuff to the actual building-it-together process that - hopefully - only takes an hour. I’d like to achieve this by having the custom wood parts milled with a computerized numerical control (CNC) machine.

Tools, tools, tools

Such a machine wouldn’t eat the plans I scetched by hand. So I have to familiarize myself with software that is able to transform 2D scetches into a 3D model, and finally another tool to translate the components of the 3D model into paths for the CNC machine to cut my parts. I chose the FreeCAD software to do the 3D work.

Every beginning is hard.

But FreeCAD… well, I’ll admit that I’m not a CAD designer or mechanical engineer. But still, this tool is hard to get started with. Luckily, there are some quite good tutorials out there like this: youtube link but to be warned, I needed a couple of hours before I had drawn my first part. Plus, what you can see on the image on the right is still a long way before I was done.

FreeCAD workflow

FreeCAD will not tell you how to best work with it. It is a very versatile tool that fulfils many purposes. As I want to use it for drawing 3D parts and/or for G-code generation, I’ll only need a handful of its workbenches.

Of course, the following lines cannot be more than a quick start guide. There are tons of helpful materials out there like a FreeCad wiki, a forum, and many, many tutorial videos etc. at your hands that explain things better and in more detail than I can ;)

Dimensions first - Spreadsheet workbench

Let’s create a central place where we keep all our project’s dimensions. We’ll link them in the drawings, 3D models, and assemblies later so that changing values in the spreadsheet will automatically propagate.

Select the “spreadsheets” workbench. Create a spreadsheet and rename it e.g. to Dimensions.
Draw a chart with your dimension’s names. I found it well readable to have a part_description_attribute naming convention for every dimension.
After entering your dimension’s value, right-click its cell and select “Properties”.
In the following dialogue, select the “Alias” tab and enter the dimensionName you entered before
In any scetch, drawing, parts design or similar workbench, you’ll be able to use these dimensions under their Alias name. Click “Save”.

Sketch it! - scetcher in Parts Design workbench

It’s time to create a first sketch of your part using the dimensions you just created. You’ll be drawing so-called master scetches fist which contain all elements of a drawing at a specific view, e.g. on the XY-plane.

Select the “Part Design” workbench.
Create a new body using the blue lego-ish symbol.
Create a new scetch for this body using the scetch symbol. Select e.g. the XY-plane.
Rename your scetch so it’s easy for you to remember its purpose. As your first scetches are the master ones, I propose to name them master_yourPartName_xy
Double-click the sketch and start drawing your shapes.
Make sure you fully constrain (youtube link) your scetch. Use the dimensions you entered in the spreadsheet to help: Double-click your dimensional constraint and type = or press the tiny round fx button. Enter the name of your spreadsheet (there is auto-completion, luckily) and then the alias of the dimension you want to constraint to reflect.
Repeat for any dimensional constraint.

With your completed master sketches (one each for every dimension), all information is available for any third party service to mill your parts. If you wanted to do that, you could export your sketches to a dxf file here and you’d be done. Else: please continue reading.

Create 3D parts - Parts Design workbench

To create an actual 3D part from your master sketches, you’ll have to create derivative sketches that can be extrudes or cutouts. You’ll have to understand FreeCAD’s way of tree-aligning parts and scetches first: Every pad has an underlying scetch that defines how it looks like. Its extruding direction directly depends on the sketch’s plane. Each part can have one or more datum planes where other parts can be attached to, again forming pads, holes, or pockets on that specific face.

FreeCAD will only display the “last” member of your part’s tree, containing all dependant parts you designed before. To toggle visibility, simply hit the space key.

Create a new sketch in your part’s tree structure. This will be your first pad, i.e. your actual 3D part.
Choose the create an edge linked to an external geometry tool (hotkey X) and select the basic pad of your originating master sketch.
Draw your shape to match this shape and set its constraints, again using the Dimensions spreadsheet you created earlier. You shouldn’t need this too often though, as you linked all possible edges you could find to your master scetch before.
Now comes the fun part: close the sketch, then hit the pad a selected sketch icon and enter your dimensions.
Done!
To create another pad, a pocket or a hole on top of one your pad’s surfaces (called “faces”), first create a new sketch and name it meaningfully. Then click the face you want this scetch to be based on. Finally, choose the “set the Support of a scetch” tool and select the scetch you created before. Repeat starting at step 2. until your part is complete.

Hm, your new pad sits right on the master scetch so you cannot see them any more? Don’t panic, just continue reading.

Arrange 3D parts - Draft workbench

In my opinion, it’s a pain to rearrange parts in the Parts Design workbench. It’s getting much easier if you use the “Draft” workbench for this task. You can clone parts easily here as well and arrange them as you deem fit.

Select the “Draft” workbench
Choose the pad/part you have been working on before in the parts tree.
Click the sheep icon to clone your part.
Make the original pad invisible by hitting space key with the part active.
Choose the cloned part and rename it meaningfully, e.g. myCoolThing_part.
Select the “move” icon (marked green in above image).

A move consists of two steps: first you have to enter the reference location for your move. Either select the point by clicking somewhere in the 3D view and then moveing the cursor a bit (my recommendation) or enter the reference in the fields on the left, confirming each axis with Enter. Second, is the actual move vs your reference point. I recommend to enter those values using your keyboard because it’s incredibly hard to move to an exact location using the mouse. Confirm with Enter, and the move should be complete.

Cool, you now have a dimensions sheet, master sketches that fully specify your parts, the parts themselves as 3D bodies and all properly that’s neatly sorted in your view!

TechDraw: Create a classic “Paper 2D” drawing as PDF

Now, maybe you want to export classic 2D drawings of your part? Luckily, that’s an easy task!

Select the TechDraw workbench
Press the Insert default page icon.
In the parts tree, select the 3D part you’d like to have drawn
Press the Insert multiple views of drawable object(s) icon and select the views you want to be present.
Add dimensions, annotations etc using the tool bar.
Use the turn view frames on/off if vertices overlap your geometries, so you can select them again.
Insert cosmetic vertices if you need to, e.g. to draw symmetry lines.

The resulting document might look like this:

Further information regarding TechDraw workbench can be found in the FreeCAD Wiki.

How to create better PCB designs

Schallbert — Tue, 21 Sep 2021

Motivation

I have my first request for a copy of AnywhereAmps Alpha. I wouldn’t want to spend another 10h in building it, so I decided to invest the time into speeding up the manufacturing process. I figured out I could save about an hour per amp if I had the circuit board produced for me. Unfortunately, switching the Circuit Board from hand cutting, milling, drilling, soldering to a PCB design cost me twice as much time compared to a full build already. And I don’t even know if this works yet. But let’s start at the beginning.

Requirements on the new PCB

The PCB shall contain

the preamp circuit as designed for AnywhereAmps Alpha
the two-way crossover circuits as designed for future AnywhereAmps Beta
a switching buck converter to provide 9V DC for the aux effects
a Power supervisory circuit protecting both battery and PCB from overload, short circuit, under voltage, reverse voltage

This way, I want to gradually professionalize the electronics. I intend to use the same PCB for any variant of AnywhereAmps. In addition, I would get rid of some external components that I had to manufacture and place by hand: Fuse, high-current capable power switch, the 9V regulator and its cooling, and some wiring.

Component selection

Buck Converter

I searched the web for a low-EMI (ElectroMagnetic Interference) switching buck converter with an output of >1A @9V. A linear regulator would have been less noisy, but at a nominal battery voltage of 18V I would have to cope with >9W of thermal losses worst case. Plus, this solution wouldn’t have been battery-saving at all. I selected a (costy) design that has all necessary components in-a-box so I just had to get the PCB layout right for EMI reduction. Add a couple of bypass capacitors and a resistor divider for the target voltage setpoint. Done! (I thought at that time. Here’s how it actually turned out)

Power Supervisor

The only protection for my amps was a fuse until now, its value set to the highest “normal” current I expected in the circuits. In the early prototype phases I killed some opamps by accidentally applying a reverse voltage. And I want to avoid to totally drain the battery pack when I forget to look at the charge indicators. So I selected a power supervisor IC that would detect any undervoltage or fault condition and protect both circuits and battery from damage.

Passive components grouping

I learned that manufacturing a PCB with many different parts would be more expensive and more prone to errors than using less different parts, even if the total number of parts increased this way. So I redesigned many values and re-did the simulation for the audio preamp with these new values. Finally, I was able to break it down from 8/5 to 3 different resistor and 3 different capacitor values!

Capacitors, again

I first used MLCC (Multilayer ceramic capacitors) in my design, both for bypass, and for the audio decoupling capacitors. After again reading the documents I pointed out in my preamp design considerations post, I realized that this capacitor design seems to have trouble with distortion on audio signals. The image below shows a comparison of different capacitor types and sizes. You see that film capacitors have the best performance and MLCCs are worst. As I wasn’t able to find film capacitors with the size I needed on a SMD footprint, and aluminium electrolytic caps were too large, I selected polarized tantalum capacitors. For the bypass uses I selected low-ESR types.

TI Audio Design Journal: Capacitor types compared ^[1]

Circuit design

I chose the open-source software KiCad as an Electronic Design Automation tool and redrew all my existing circuit plans, adding the components described above. Look how the amplifier section only fills the right half of the schematics while the left half is devoted to power supply and supervisory circuits.

I tried to get the schematics as neat as possible to make them easy to read. I used global labels a lot and added connectors for simplified mounting.

PCB design

This was the hardest part for me. I never designed a PCB for audio usage before, so I had to read a lot of guidelines before I started.

How tos and tutorials

I list some of them below for your reference:

noise on PCBs and how to avoid
how to best place bypass capacitors
ground plane or not?
how to handle different grounds for signals, power, etc.
how to optimize parts placement for soldering
avoid these pitfalls when designing PCBs
how much current can my track handle with this width?
some PCB design specifications for manufacturability I tried to apply those design guidelines the best I could.

AnywhereAmps PCB explained

The result after a couple of evenings:

I separated the PCB into three zones.

Left: the 9V switching regulator, following its reference design (datasheet) it would create a current circle around itself using the bypass capacitors and cancel out most of the EMI. Look at the thermal design, it has a lot of copper and quite some vias to get the heat conducted away from the die.
Center: The power terminals and supervisory circuit. I placed solder jumpers to select the nominal battery voltage for the undervoltage detection.
Right and top: Preamp and analog signal circuitry, separated by the other electronics stuff using a thick ground track on the top copper and an own ground plane on the bottom. I hope this is enough…

As I’ll hand-solder the first prototypes to check if it all works OK, I used “big” 0805 SMD components. I increased the space in between them by more than the PCB specification would demand. Unfortunately, the highly integrated power ICs were not available in “hobby soldering electronics” sizes so I’ll have to go through some pain there :(

So this is how it will look like. The PCB size will be 45mm x 69mm.

Contribute or download

All the CAD models and schematics are open source and available for download and your contributions on my dedicated Github repository.

Up next: PCB delivered and soldered

Preamp design considerations

Schallbert — Wed, 25 Aug 2021

Decoupling capacitor selection

Some tech deep-dive background information about how I selected components for my preamp circuit board.

Why decoupling is needed

By decoupling, I specifically mean direct current (DC) decoupling in this case. The preamp needs to be decoupled both from the source (the musical instrument) and the sink (the power amplifier). This has multiple reasons:

the supply voltages of the three partners may differ (e.g. 9V guitar phantom, 18V preamp, 52V power amp) and I don’t want to accidentally short-circuit one of them
Output impedance, I mean the load capacity of the output stage might be limited. If I connect a low input resistance e.g. directly to a guitar’s pickup, I might not get any signal at all because the source is too weak to power it correctly.
Noise reduction through electric coupling

Why capacitors can do the job

There are multiple ways of electrically decoupling circuits. You can use a photocoupler to totally (i.e. galvanically) separate two circuits from each other while only the signal is passed between the circuits, or - in my case - a capacitor to only let alternating currents pass, and to block constant current altogether.

How does this work?, you ask? Well, a capacitor acts a bit like a water tank. If you apply a constant voltage (pressure), the tank will flood to a certain degree. When this level is reached, there’s no flow (current) anymore, neither in nor out. But if you now vary the voltage around this static level (water pressure variation), you will measure some medium (current) flowing in and out of the tank again.

It’s the same with the capacitor. If you connect it in series to a sink, it will only pass current over when the input voltage level is changing.

Decoupling Capacitor type

Multilayer Ceramic, Polypropylene, Electrolytic…? After reading this guide from Texas Instruments, I understood this document is all about distortion. Distortion would hit the hardest on the instrument input before the preamp, because this will be amplified by both amp stages. When using an electrolytic capacitor, there should be a voltage bias applied to reduce distortion in the low frequency range. With my design, the bias voltage is 1/2 of VCC, so even higher than the 5V used in the guide - I think, I’ll be fine with a “cheap” polarized Aluminium cap on the THD+n side.

There’s another thing about MLCCs I noticed myself: With loudspeakers disconnected at high preamp gain, the power amps I’m using still play music! How’s that? The power amp board is using MLCCs which act like piezo actuators. Their vibration is amplified by the PCB they are mounted on which gives a proper sound level. Crazy. For more geeky details, have a look at EDN’s blog post or - again a Texas Instruments guide.

Capacitor size

To judge capacitor size, I had to take a couple of aspects into account:

A bigger capacitor will lower my bottom end cutoff frequency
A bigger capacitor will put more load on the input source due to charge losses
A bigger capacitor will cost more, especially the high-quality ones.
A bigger capacitor might have a larger footprint.

I wanted a lower cutoff of <20Hz for full HiFi capability. As that’s the -3dB point already, I chose it to be even lower at 15Hz calculated via \(f=\frac{1}{2 \pi RC}\). With an input impedance of 10kOhm (the value of the resistor clamping to virtual ground) I yield a target input capacitor size of 1µF. With this capacitor, the input source will be considerably more affected by the 10kOhms input resistor than by the capacitor with its low series resistance (ESR) of some hundred mOhms, so the chosen value is fine.

For the output capacitor, the calculation is similar. The power stage has an input impedance of 20kOhm, so the lower cutoff frequency of 17Hz is reached wich a 0.47uF capacitor already. Anyways, to keep value differences at bay, I decided to go with the 1µF as well.

Back to Project main page

Shelly13 - Part2

Schallbert — Fri, 09 Jul 2021

Shelly13: Project status, prototype scetches

I’m still thinking a lot about Shelly13 and how it can possibly work out to have a bigger speaker, a homogenous sound with some punch, and a sealed cabinet design that can be folded.

Cabinet: Sizing and Volume

Latest designs in my head work with 12x10 and a 14x10 inch shells. It will look a bit like a cake when unfolded, with a total cabinet volume of around 38l incl. speakers and amplifier mounted inside. It will be about 75cm of total height with its feet unfolded and about 29cm high when completely folded.

Speaker(s)

While my first design shows a 10" speaker mounted at the front of the cabinet, I’ll most likely switch to a two-way design for simplified mounting, mitigating the issues about milling a huge curved surface and creating an adapter plate. So I’ll chose an 8" Bass speaker in a downfire setup and a 3" mid-high speaker at the front.

Amplification

I’m thinking about using a car amplifier that takes its signal from an active crossover with separated outputs for low and high frequencies, creating a bi-amp design with bridged terminals.

Folding mechanism

Folding will take place with the help of drawer rails mounted in between the shells. It is crucial that the cabinet is sealed air-tight when unfolded, so there has to be a ring gasket that engages when the lower wooden ring at the bottom of the 12" shell meets the top ring of the 14" shell.

Nah, this is too complex!

Ok, let’s strip it down. Leave the foldability away for the initial prototype. Reduce cabinet size instead. Concentrate on electronics and manufacurability first. Outcome of this scope change is: AnywhereAmps Alpha!

Shelly13 - Part1

Schallbert — Mon, 28 Jun 2021

After playing MobFobAmp for a while, I wanted to take another step in the evolution towards a really good portable instrument combo. With this design I’m trying to address the shortcomings of the previous prototypes:

Power output could be higher
More volume in the low lows would be nice
Folding mechanism a bit fiddly.
I’d like a one-step unfold, preferably one-handed operation possible
Mechanical stability and robustness could be improved
Sensitive components mounted on the outside requiring to be handled with care

A first design

Taking the above points into account, I figured out that I had to compromise foldability a bit to get all the sensitive components including the speaker chassis to the inside of the cabinet. A ball-shaped design would be ideal for best low frequency / size ratio, but I didn’t know how to manufacture such a design - so I went with a cylindrical approach. I’d like the speaker to “pop up” so it would be protected from mechanical damage when folded. So my first idea looks like this:

Project constraints

I cannot manufacture the cylinders myself so I’d have to contact a drums manufacturer to send some unfinished custom-size Tom-toms over to me. Even if I get those, I’ll have to cut on a curved surface which I never did before, and I’d require some 3D printing / routing to craft an adaptor plate between speaker and cylinder… Or will I just have the speaker mounted on the top/bottom? Plus, I’d have to think about how to include a more powerful amplifier into my design - there are some ideas like this Pioneer car amp or the specialized Hi-Fi amp from DIY Class D but they have very different and maybe harsh requirements on power supply…

Conclusion

You see, many open items to think about. I’ll let this sink in for some time before continuing work on this project.

AnywhereAmps - Alpha

Schallbert — Mon, 01 Jan 0001

Mobile Instrument Amplifier/Cabinet combo prototype based on Shelly13. AnywhereAmps Alpha has a low power, small footprint aproach. It is designed as a very portable full frequency range practice amplifier with a total weight of only 3.1kg. It features adjustable gain and a 9V DC supply for external effects pedals.

Project stats

Difficulty: medium 3/5
Cost: ~170€
Time: 10h

Impressions

Impressions of AnywhereAmps Alpha's prototyping

Can be played upright, especially suited for very small practice rooms.

9V voltage converter for floor pedals and accessories.

Minimalistic HMI: On/off switch, gain, 1/4'' jack, 9V out

12'' diameter, 2.6Gallons of volume

Battery on the back plate. Pull up to remove.

Preamp and poweramp are stacked on top of each other.

Amp sound samples

Bass samples

Your browser does not support the audio element.

Predominantly high notes

Your browser does not support the audio element.

Played really hard

Your browser does not support the audio element.

Open strings & flageolet notes

All tracks recorded with Zoom’s H4n using an Ibanez SR1500PD from 1992 with the neck pickups @100%, full volume, low string setting.

Dimensions

Raw drum shell: 12"x6". Total height is around 17cm, width 30cm, depth (without battery) at 32cm. The enclosed volume incl. amp and speakers is ~9l. I’m using a non-vented cabinet design.

Sound Hardware

Speakers: 4" Bass/Mid and two Piezo tweeters
Class D amplifier with 30W RMS power (bridged)
14.4V Li-Ion tool battery supply
All sensitive parts on the inside of the cabinet
Custom built preamp with 0...27dB adjustable gain

Simulation

The simulation shows that the volume of ~9l damped cabinet would be a perfect match for the selected 4" speaker. It will go down as low as 44Hz (F#1, one full note above e.g. a Bass’s lowest open string E1). There will be punchy lows at a precision which is still good. The downside is the low efficiency of just 78dB/W/m, so the expected maximum volume I’ll get from the power input will be rather low and the speaker will be driven into its excursion limits early.

Custom Preamp

The class-D amplifier I selected requires a line-level input, so I had to add a preamp to my design to get sufficient volume from the power stage. Even the cheapest Preamps like this one from Art Pro Audio would add considerable cost to the design. They would weigh a lot and consume quite some power, especially if they are tube preamps, and often have special requirements on input voltage (like AC only). Plus, their design wouldn’t blend into mine.

So I decided to design a low power, minimal parts gain stage myself. Click here to skip the nerdy Tech details

Voltage rails

As the preamp would amplify alternating voltage signals but I only have one battery voltage rail available, first I had to think about creating a virtual ground potential that would be the point of signal zero for the gain stage. Luckily, there are a lot of design guidelines about how to operate the amplifier with a single voltage source, so I had something to start with. I used 10kOhm resistors to clamp the inputs of the gain stage to the virtual ground.

Gain stage

AnywhereAmps Alpha should be a multi-purpose instrument combo, so that there should be a wide gain range available to bring everything from a passive guitar to a line level Keyboard to the same voltage levels for the power stage to deal with. That’s why I decided to go for a 0-27dB amplification range and selected resistors for the feedback loop to be 10kOhm and a 250kOhm logarithmic potentiometer.

Circuit simulation

I used Analog Device’s LTspice to simulate how this circuit might behave when given a typical Audio signal between 20Hz and 20kHz. From my studies I remembered that I might be most interested in a so-called Bode plot which would list both amplitude and phase over frequency in a double-logarithmic scale. Conveniently, LTspice offers such a function. As you can see, the preamp’s output - green - stays flat into the 10kHz range and the phase would only shift a little while the input line -blue - stays ~27dB below.

Op-Amp selection

Well. I tried this design on my breadboard with multi-purpose Rail-to-Rail Operational amplifier I had lying around and it worked… not well at all. When I connected a function generator at the input and measured what was supposed to be a sine wave at the output of the LM324, I didn’t get a sine wave at the output at all, but a half-sine with some steps and discharge-alike curves around 0, then followed by the negative half-sine… After some search I found out this is called “Crossover distortion” and is caused by the asymmetric output stage of this specific op amp type. I was able to correct this a bit when applying a higher load current on the output, but finally I wanted to choose an op amp which was better suited for my audio application without rail-to-rail capability and lower noise ratios. Thankfully, there are quite some pages on that topic like Cicfy Research’s Op-Amp shootout. Weighing in a reasonable piece price and the availability at my electronics distributor, I selected the NE5532P. The video below shows what I got from the updated op-amp selection.

Your browser does not support the video tag.

Preamping a Sine wave

Circuit design

I wanted my circuit board to use as little space as possible while keeping DC voltage supply panes like a frame on the outsides of the board. The board’s left half is dedicated to creating the virtual ground potential while the right half does the amplification part. The design is minimal, using only as much as four resistors and three capacitors.

For a technical deep dive, refer to my design considerations.

Sound quality

AnywhereAmps Alpha is absolutely quiet in idle, even with gain full up there’s only so much as a little hissing from the tweeters. No hum, no scratching when turning the gain knob. It works with any instrument due to the wide gain range using 0dB for line levels like Keyboards, preamped instruments, mobile phone, and up to +27dB for passive Bass or Guitar direct in. Maximum volume is fine although the low-lows “flatten down” too early, limiting usable volume range e.g. for organs and 5-string basses. For frequencies in the Bass range at high volume, it still comes down to Physics - the bigger the better. Find some sound samples up here.

Summary

For the cabinet, I ordered some raw custom size Tom-tom shells from Cube Personal Drums. Cutting and especially hand-routing a curved surface proved to be hard, and the fact that I don’t own a proper wood workshop and did all this work on the balcony of my flat didn’t make it easier. But after roughly 10h incl. soldering and electronics work, I now have more than a super-portable instrument combo.

It’s a piece of furniture.

I’m going to be playing it a lot. Plus, the good sound keeps me more motivated than ever to go the next step and boost reproduction of the low frequency range with another, larger prototype.

Pros:

I like the design.
Fits into standard Drum bags!
Lightweight, 3.1kg incl. Battery, 2.6kg without
Battery state of charge can be read at a glance
Retracted controls: Avoids inadvertent operation
More robust: All electronics and speaker chassis mounted inside
includes 9V, 1.6A power supply jack for effects, stomp pedals, etc.
Sounds good for any instrument as it plays almost the full audible frequency spectrum
Runs >20h on a battery charge (14.4V, 4Ah) as a practising amplifier and moderate volume
Runs >3h on a battery charge at maximum volume

Imperfections:

Too small for sub-bass reproduction.
No auto mute on switching the amp on or off -> Pop Noise
Of course, not enough volume to match a drumset or plugged band setup.
Looks weird with big battery packs on the back. Use of a “flat pack” would be nice

AnywhereAmps Alpha: DIY Replica

You want do build your own copy? Sure, use the step-by-step guide below and get going! AnywhereAmps Alpha DIY instructions

Your browser does not support the video tag.

Alpha live!

AnywhereAmps - Alpha 1.1

Schallbert — Mon, 01 Jan 0001

Mobile Instrument Amplifier/Cabinet combo prototype Version 1.1, evolving from AnywhereAmps Alpha.

Project stats

Difficulty: medium 3/5
Cost: ~200€ plus battery system
Time: 3h

Updates over Alpha 1.0:

Add power supervisor circuitry to protect both amplifier and battery pack from abnormal voltage levels / currents
Integrate on/off switch in potentiometer
Reduce complexity by using high efficiency full-range speaker instead of 3-speaker 2-way system
Add high efficiency switching buck converter to generate 9V effects voltage rail
Save construction time by using custom printed circuit board featuring preamp, step-down converter, and power supervisor
Save construction time by delegating most of the woodworks to a portal milling machine CNC¹
Eliminate switch-on pop noise by adding a RC-filter to the standby input of the power amp stage

Build variants

To get progress on sound for my prototyping, I built different variants this time. I changed speakers and evaluated the sound:

Alpha 1.0: 30W bridged amp, 4" bass and 2x piezo tweeters: HiFi-esque sound with punchy lows and clear highs, limited on the max volume side
Alpha 1.1 v1: 30W bridged amp, 4" PA driver, good volume and high presence, but underrepresented lows and highs
Alpha 1.1 v2: 30W bridged amp, 4" PA driver and passive membrane, fixing low-end

My own preamp and auxiliary functions printed circuit board

The AnywhereAmps 1.1 PCB features some extra functionality apart from the mere preamp plus lowpass and highpass for bi-amping. The details can be found in the following sections.

Power Supervisor

I added the power supervisor circuitry to not allow the battery to run completely empty. I also wanted to mitigate risks created by short circuits and other fault conditions which could harm the battery. The integration of this technology worked on first try (which I hadn’t expected). The only drawback is that when the battery gets discharged to below, say 10%, the amp would sometimes restart on power bursts as supply voltage would drop below the supervisor’s trigger voltage. As the battery voltage would go up again when load reduces, it will turn back on a second later.

This is clearly undesired behavior and will have to be handled with a refined prototype level. Another thing that makes this follow-up prototype necessary is the fact that I am not allowed to use the tool manufacturer’s batteries in my future designs as discussed here.

9V switching buck converter

To generate auxiliary 9V for devices like effects, loopers etc. I wanted to reduce noise by selecting a “quiet” switching regulator. The one I chose was designed for automotive applications and has the inductors included in its enclosure. The down sides are that the footprint is not suitable for hand soldering and the high price tag.

The converter would switch on when a device is plugged into the corresponding jack so that there’s no stand-by current. It operates quitely (but not completely without noise) and more than enough power for the applications I encountered so far - A looper, compressor, and tuner can be easily operated in parallel.

Woodworking

I bought a portal milling machine set to delegate the woodworking for AnywhereAmps to. I wanted to spend less time on cutting and sanding, and improve build quality as well as reapeatability. But changing the manual woodworks to more or less automated machining, I had to first move my plans into the digital world.

Computer Aided Design

I chose FreeCAD as design tool to create drawings. I documented the work with this tool in a couple of posts and continued to refine my workflow e.g. here. My first scetch to manufacture looked like this.

I was not completely happy with partitioning the sheet of wood yet. I found the legs too long and put some more hours in refining the design to use less space as the handling capacity of my portal milling machine is limited.

Computer Aided Manufacturing

When done, I exported the scetches as *.dxf files (Drawing Interchange File Format (DXF), introduced by AutoDesk in 1982. It eventually became industry standard for interchanging 2D drawings) and opened them in my CAM tool to define tool paths, set XY zero, and Z depth per pass.

G-Code

I ended up with G-Code files post-processed to match my machine with a forecast duration of 32 minutes per sheet. I was using just three tools: a 6mm endmill for the cutouts, a 2mm one for smaller holes and structures, and an engraving tool for chamfers and logo. I partitioned the work into three jobs:

job00_clamping to cut holes and countersinks into the sheet to allow attachment to the workbed
job01_nesting to do all cutouts and engravings on the ampcarrier, but leave taps so the parts don’t move
job02_finishing which cuts the logo on the (already coloured) top side and adds chamfers where needed

For finishing, I had to update CAD and CAM files so they would provide a mirrored (and turned by -90°) view of the sheet. This is to allow two-sided milling with the same XY-zero point and the exact same sides touching the fence for minimal tolerances.

Manual work remaining

I had to do a little manual work like sanding down taps, adding wax to get the top finish done, and cutting some holes for struts and stands that my CNC couldn’t do because they are on a plane orthogonal to the Z-axis.

I used the CNC to create adapters for my hand router so I could have it glide over the bent surface of the shell. This way, the holes that I cut in there for the speakers before could be properly chamfered.

I also had the CNC manufacture a shell holder in 3D-mode (with *.STL files). This way, the shell is kept in place when cutting holes for the speakers with my hand drill. By the way: 3D cuts take a lot of time as finishing uses small stepovers to get a smooth surface. I’d like to further improve shell handling; hand-cutting a >100mm hole does not allow good surface quality on the inside of the shell, and handling that huge drill is still a pain. Unfortunately, my CNC is not able to to this job for me.

Impressions

Impressions of AnywhereAmps Alpha V1.1

Battery attached and switched on

Front view, initial version of speaker protective grille

A view from below shows the passive radiator, extending the speaker's low-end

Preparing final assembly

A view inside: Ampcarrier and circuits

Allowing the wax to dry before final assembly

Computerized Numerical Control ↩︎

AnywhereAmps - Alpha 1.2

Schallbert — Mon, 01 Jan 0001

Alpha 1.2 v1: 2x15W bi-amped, 2x3.2" mid-high speakers, 8" subwoofer: very good low and mid, high highs a bit dull, weighs a lot
Alpha 1.2 v2: 2x15W bi-amped, 4" PA driver, 5.5" subwoofer, 2x piezo tweeters: Very balanced sound, build is pretty complex.

I’ll continue writing here once I selected the best variant.

CarSubAmp

Schallbert — Mon, 01 Jan 0001

project stats

Difficulty: beginner 1/5
Cost: ~50€ without battery pack
Time: ~2h

The CarSubAmp was my first prototype for a really portable Bass Combo (Speaker cabinet + amplifier) that doesn’t even need a wall socket.

Motivation

During Covid19 Pandemic, I enjoyed playing and excercising Bass outside in the Parks or on my balcony. Which is not a problem when wearing headphones and carrying a small preamp. But, there were occasions when I wanted to play along with, say, an Acoustic Guitar, or somebody would bring a Synth-App along with a MIDI Keyboard - and I was forced to use a small bluetooth speaker that neither had the output power nor the real-time feel.

Problem

My Bass amp is far from mobile. Weight: 18kg, size: huge, plus: It’s useless without a wall socket nearby.

Dead ends

Instead of a Class-D amp I used an old Car Radio’s aux in, but it seems to have expected “Line” level which my bass couldn’t supply so the output was rarely audible. Additionally, the idle current draw was >0.5A so it would drain the battery all too quick.
I tried the power supply with a powerbank (Power-Delivery) first, but the integrated step-up converter didn’t seem to handle full-volume current surges, so the supply voltage dropped, leading to audible distortion.

Solution

After some thinking, I came up with the idea of using a cordless multi tool’s Li-Ion battery that would be able to handle current surges easily, have a low weight, feature a battery level indicator and the typical protection circuits to avoid overload, short circuit, over temperature, exhaustive discharge etc. I wanted a Class-D power amplifier matched to the voltage range of this battery to be as energy efficient and as lightweight as possible.

The “CarSubAmp”

To check if the concept was viable, I took a used OEM car subwoofer assembly that normally goes into the spare wheel well, bought a 20W amplifier developer board and after soldering some components, adding a fuse and an 6.3mm input jack, my first prototype was ready to try.

Project summary

Very lightweight, <5kg
4Ohms Subwoofer assembly “ready to use”
vented design for better speaker efficiency
Sound is OK, but it’s clear that the assembly is not meant to be used as a full-frequency range speaker. Muffled.
Playing low notes (i.e. at or below g1), the system would become noticeably quieter
Developer board comes without preamp, thus maximum volume a bit low already
Under high load, vented design “Bassreflex” leads to audible hissing noise
Due to its small diameter, the speaker would hit maximum excursion limits when playing Hammer-ons on low strings
All components fixed with hot melt or tape, thus not very durable
Assembly is sealed with glue, no way of permanently adding a case or enclosure for battery, amp, cables, or input jacks

Altough this design had clear disadvantages, I actually enjoyed it a lot because for the first time, I could literally play anywhere and anytime I wanted. I even caught myself on the couch with this little amp although the “real” Amp was just one room away.

After some months carrying this thing around, 3 re-solderings of broken connections and glueing the battery back into place again and again, I decided not to stop with this prototype.

MobFobAmp

Schallbert — Mon, 01 Jan 0001

A second prototype of my mobile Instrument Amplifier/Cabinet combo with foldable design. 28l cabinet with 17cm woofer. 15W amplifier power. 7.3kg incl. battery. Especially suitable for Bass Guitar, Acoustic guitar, Organs.

Project stats

Difficulty: challenging 4/5
Cost: ~140€ without battery pack
Time: ~15h

Impressions

Some images of my fold-flat bass combo 'MobFobAmp'.

Consumes very little space when folded

Easy to carry, weight is less than 8kg.

28l of volume when unfolded.

Height of 20 inches, can be used as a stool.

6.5 inch low-end driver, 4Ah battery pack for hours of battery life.

A peek inside

Your browser does not support the audio element.

MobFobAmp Sound sample

Your browser does not support the audio element.

A/B/A comparison Sealed vs. Vented

Your browser does not support the audio element.

MobFobAmp and Piano, Sealed, Preamped, at maximum volume

All of these tracks have been recorded with my Smartphone, so no high-end gear used. Still, I think they represent the actual sound of MobFobAmp quite well.¹

“MobFoBamp”: Motivation

To overcome many of the constraints of my previous prototype, I decided I’d need a bigger speaker, and a more voluminous enclosure to yield a better response on lower frequencies. Unfortunately, a bigger enclosure would considerably reduce mobility, so maybe I could get rid of all the air inside the box for carrying…?

The Idea: Folding box

A folding box ^[1]

Following the concept of a folding box like shown, I decided my combo should be collapsible for easy carrying. I chose a sealed design for starters so I wouldn’t have to care about calculating vents, resonance frequencies and group delays.

I had to think hard how to integrate speakers and electronics while keeping the ability to fully fold the cabinet.

Dimensions

I selected the cabinet’s width in a way that it would fit snugly into my bass gig-bag if needed. All other dimensions fell into place by itself - height had to be at least twice the width so top and bottom lids could be accomodated when folded (but not too high to keep stiffness as high as possible. Depth had to be less than (0.5* width) - (materialStrength + hingeThickness), otherwise folding wouldn’t work as the sides would get in their way.

Sound Hardware

For the power Amplifier I chose to stick with the module I already bought earlier. I chose a 6.5inch Hi-Fi woofer (50W, 4Ohm) with low resonance frequency to match my sealed cabinet, which I figured would have a volume of ~28l.

Simulation

Then I ran some simulations to see whether cabinet design and speaker would be a good match.

Estimated Qtc would be at 0.63 instead of the recommended 0.71 (optimal step response), so the cabinet would be somewhat overdamped and sound “tighter” but loose volume at the low end compared to an optimal setting. As you can see, building a cabinet always is a compromise between sound, power, size, weight, frequency response, and volume level that can be obtained from that setting.

Summary

When I finished building that cabinet after some 15h, I was quite happy to see the cabinet is airtight and not flimsy at all. Unfolding would take me below 60 seconds, folding it back to its original state even less than 30 seconds. As maximum level still was a bit low, I added a compressor stomp pedal to the setup which had a gain knob I could use as a preamp.

The sound of this thing is amazing. Tight Bass, warm mids and subtle, but brilliant highs fitting the characterisitc sound of the bass guitar very well. Maximum sound level is limited, but it’s always enough to keep up with an acoustic guitar. I’m using it as a practising amp all the time.

Sound quality

For sure, the most important aspect of an instrument combo is how it sounds like. Due to the foldable design, it can be operated in two different setups.

Sealed: Tight and well-balanced sound, good low-end punch. Without preamp, maximum level OK to play along with an Acoustic Guitar
Vented: Bottom Lid opened ~3cm, latch suspending cabinet against bottom panel. Much louder sound with pronounced low-mids, less balanced and a bit boomy, maximum level on par with Piano played f.

I’m really happy with the sound quality MobFobAmp provides. The option to play “open” adds flexibility and makes the sound really stand out (with the drawbacks mentioned above).

I also have performed some measurements to see how close the design gets to an ideal frequency response, sending some White Noise to MobFobAmp and then recording the sound output with a microphone. For comparison, I displayed a spectrum analysis in the frequency domain. A perfectly neutral sound output would hit the orange line at 0dB for the whole spectrum.

You can see, that the -3dB limits or useable frequency range is from 30Hz to ~1.5kHz.

Add-Ons

9V power supply for Pedal Board

I’m using some effects pedals when practising, and I also wanted to be able to use them when I’m taking MobFobAmp with me. That’s why I decided that it might be a good idea to have a +9V supply rail with barrel connector (barrel = +) on board. So I bought a very well-known positive linear voltage regulator from the 78xx series called 7809 in a TO-220 design that can supply up to 1.5A of current at 9V, and accepts an input voltage range of 11-35V. The component can be used with minimal additional parts as shown in the image on the right. This way, I can now drive a pedal board from the battery of MobFobAmp!

Piezo Tweeter

When I first used a drum computer with mobfobamp, I felt that the highs just were not enough. While perfect for an electric Bass, the drum machine’s hi-hat, cymbals, and even the snare sounds dull and mumbly. So I ordered a simple piezo driver which I could connect in parallel to the main speaker - this works perfectly well. I added a 2W/320Ohm resistor in series with the driver to get the levels just right to my ears. The result is great TODO: and can be heard here.

MobFobAmp Pros:

Foldable to very compact dimensions
Low Weight: 7.3kg incl. Battery, 6.9kg without
Runs >20h on a battery charge as a practising amplifier and moderate volume
Runs >3h on a battery charge at maximum volume
Sounds really good for Bass, Organ or Acoustic Guitar
Playing low notes down to B0 works fine and sounds punchy
9V DC out for preamps or effects boards
Can be operated in “Bass Reflex Mode” for more Bass

Imperfections:

Without Piezo tweeter, highs are muffled
Low power amp headroom: When turning up the external preamp too much, the power amp will distort notes, especially low ones, first with flutter, then with the typical sound of transistor saturation
Prototype’s rat’s nest wiring is still prone to mechanical damage, so an additional enclosure for the electronics would be nice
Of course, not enough volume/ power to match a drumset or plugged band setup.
Protective grill for speaker would be nice

MobFobAmp - DIY Replica

If you like this idea, feel free to build your own copy! You’ll learn what it takes below. Mobfobamp DIY instructions

Piano by Feliks Weber ↩︎