Daisy Bud - a headphone board for Daisy Seed

[BJG145]

The other day I thought of a PCB I wanted to make, and I thought I'd blog the process of getting this done.

I've been looking for a way to approach embedded audio for the Gullsonix range...aka the DIY MIDI controllers I've been working on. I've thought for some time that Daisy Seed by Electrosmith was probably the best option, but the coding knowledge was beyond me.

However...there's an easier way to get started with Daisy programming, via "DaisyDuino", which adopts an Arduino-like approach. This isn't as flexible or efficient as coding directly in C++, but it's a way in, and probably good enough for me.

To make it even more accessible, there's "Synthux Academy", with original Daisy hardware, code and tutorials aimed at beginners. After picking up a Simple Touch on eBay recently...and then being gifted a year's free subscription by the nice people at Synthux...I decided to have a go with Daisy for the first attempt at embedded audio.

Now, Daisy Seed is just a chip. And there are boards you can plug it into like Electrosmith's Pod, or the Simple Touch by Synthux, but these have a load of stuff I don't want...additional pots, sliders, buttons and I/O that just gets in the way.


I just a want a headphone socket, basically. I just want to be able to plug a pair of earbuds into my Midihex controller like you can with a Dualo or a Striso.

So I decided to create my own board called the Daisy Bud. This would present a 3.5mm stereo headphone socket and Daisy's USB connector outside the case. On the inside, it would have an 8x2 pin header for connections to things like Teensy, power, pots/buttons, anything else I might want to add later...and most importantly, a serial MIDI connection, so the Teensy can play it like a soundcard.

(The Daisy might be powerful enough to do everything on its own, but it's conceptually simpler and more reliable for me to use Teensy to read the controls and just use Daisy for synth playback. Also, the Teensy 4.1 in this controller is continuously scanning 103 Hall-effect keys and using the results to make note velocity calculations, so it's already quite busy.)

I contacted an electronics designer on Fiverr who I've worked with on a couple of previous projects, and described what I wanted in simple, vague terms; pretty much what I've written here. I also included a highly detailed technical diagram that they requested.



The Daisy has line-out built in, but I wanted a headphone circuit, and suggested copying the one from the Pod.

https://daisy.nyc3.cdn.digitaloceanspac ... d_Rev5.pdf

I'm planning to get these manufactured by JLCPCB, but they pointed out that the TPA6110 used in that design wasn't available for assembly...



https://jlcpcb.com/parts/componentSearc ... xt=TPA6110

...but they identified an alternative, the TPA6111A2DR, which they said they could use instead.

I said, great! Oh, and can you include a MicroSD card reader like the Pod. And add this gull logo.

"No problem."

Cost was £80. Currently waiting to see what they come up with. As you can see, being an audio tech entrepreneur these days is pretty tough.

[BJG145]

The designer came back with the following schematic, so we're going ahead with that...

[James Perrett]

A couple of thoughts...

Is 100uF too low for the output capacitors? If you want to produce deep bass into low impedance headphones then you may need something higher. The same goes for the input capacitors to the headphone amp.

I would also suggest adding more decoupling on the power supplies. The capacitors that you have are fine for capacitors next to each chip but I think you could probably do with something around 10uF or maybe more near the amplifier.

Edit to add - see section 10.2.2.3 of the chip datasheet which agrees with my de-coupling suggestion above.

It looks like your designer has assumed that you know what you are doing electronically and hasn't checked your sketches. If you don't know what you are doing electronically then you need to work with a designer who is more pro-active and who understands audio design a little better.

[BJG145]

Thanks for the feedback James! This is an experiment, so I'll let it run...it's as much about the process as the product...but your ideas will be valuable for debugging/refining the results.

(I know nothing about audio design, and made this clear; all I provided was the above scribble...I was kidding about this being a technical doc. The previous circuit diagram was taken from the schematics for the Daisy Pod that this idea is based on.)

[James Perrett]

Looking a bit further, it appears that the Daisy Pod is relying on the decoupling capacitors on the Daisy Seed board and also blindly copying the headphone amp diagram in the chip application notes without necessarily thinking about whether the component values are appropriate for a musical device.

It will work but maybe it could be better...

[BJG145]

Just to clarify, Electrosmith created the Seed, they also created the Pod, I assumed they knew what they were doing, and asked for a design based on this. That's my fault and nobody else's. There may be room for improvement, but that's nothing to do with the designer.

Just saying this in case they happen to find the thread.

I'm glad to know you think it will work though...

[BJG145]

Layout now done. (A lot of this stuff wasn't in the original spec, but I invited them to add anything else they thought might come in handy.)

[Folderol]

Most headphones are pretty resistive rather than inductive, so if you can afford to lose a bit of volume, you can improve the bass response by adding series resistors - the higher the resistance the capacitors see, the less of a high pass filter the whole setup becomes - also gives you a degree of S/C protection.

[BJG145]

Thanks Folderol. I'm aware there's a number of electronics/audio experts on this forum (!) so I thought there might be some useful feedback.

This is an experiment to see what it's like for a novice to create an audio device, so I'm going to follow through with the original design for the first version, but I'm anticipating a v2 further down the line where it will be interesting to incorporate these ideas (and maybe improve on Electrosmith's original design).

(I'll be planning to use this with earbuds mainly.)

[BJG145]

Part 2 - Manufacturing

The Fiverr part is now "delivered", with the following files provided...

- Schematic, see above. This is done early before the PCB design starts.
- Gerber, BOM, Pick-and-place. These three allow you to order the design from JLCPCB.
- Project file. This isn't always included, so you'll need to negotiate in advance. In this case, the project was done in EasyEDA Pro.
- A couple of 3D renders (see above).

Once you have these, you can proceed to JLCPCB and upload them. I'm going to start by uploading the Gerber and accept all the defaults. Minimum PCB quantity is 5.



At the bottom of this screen, I'm going to select the "Assembly" option for the surface-mount components. I asked for a single-sided design, which is less expensive, since I only have to choose "Assemble top side". (The original Pod has certain components like the card slot on the underneath.)

The minimum assembled PCB quantity is 2, but I'm going to assemble all five. (Otherwise you get your two assembled ones and three blank PCBs.)

I'm also going to choose "Confirm parts placement".



On the next page you upload your BOM (bill of materials) and CPL (component placement list, aka pick-and-place) files, and hit "Process".

At the bottom there's a warning that no part has been selected for one of the components. After checking with the designer, I can click "Search" and select the relevant entry.



For another component, there's a warning that they don't have enough stock available, so the designer suggests an alternative I can search for.

(It's worth giving a tip when you complete the order to cover this kind of follow-up support.)



After completing this section, you can preview the model...zoom in/out, rotate in 3D.



Cost of five assembled PCBs including delivery, taxes etc...£61. (I also used a couple of coupons I'd received, worth a few dollars, from previous orders.)

Once they've checked the design, they'll send me an email to confirm the layout, then I should receive it in a week or so.

[James Perrett]

I'd review those options - you won't be legally able to sell them with the current options as they don't comply with ROHS.

[BJG145]

OK! Thanks for pointing that out...

Sounds like the lead-free HASL finish (or ENIG) is the way to go in future.



"Lead-free HASL is variant from HASL, which is lead-free and qualified for ROHS compliance products."

https://jlcpcb.com/quote/pcbOrderFaq/PC ... e%20Finish

"If pcb surface finish with lead, SMT will be with lead, if pcb is lead-free, JLCPCB will follow up the components properties to determine whether to choose leaded or lead-free."

https://jlcpcb.com/help/article/pcb-assembly-faqs

Sounds a bit vague. I've asked them about it.

[James Perrett]

You can't mix leaded and lead free parts so you need to make sure that all the parts you specify are lead free. It isn't a quick job specifying parts for a board. If you want to be sure that it performs as expected then you have to go through all the data sheets and make sure that there are no gotchas.

That's why, on a decent supplier's website, you will find a link to the full datasheet for each part linked from the part description.

[BJG145]

Part 3 - Testing

The PCBs arrived today, looking nice.





I used Electrosmith's web programmer to flash the Seed with a simple white noise generator. At first I thought the headphone socket wasn't working, then it began to cut in and out. I eventually realised that it worked when I touched the four-pin header labelled Audio/Headphone out.



Not sure what that's about...I'd assumed it was a separate set of outputs, but maybe it's a volume control, which the original Pod had but the Daisy Bud doesn't. Dunno...just dropped a note to the designer. Pleased that it's working anyway.

MIDI testing next...

[James Perrett]

Yes - that looks like a header to connect a volume control. If you don't want a volume control then just add jumpers between the audio_out pins and the respective headphone_wiper pins.

[BJG145]

OK! Thanks James.

[James Perrett]

Is 100uF too low for the output capacitors? If you want to produce deep bass into low impedance headphones then you may need something higher. The same goes for the input capacitors to the headphone amp.

I thought I'd add that I've just started reading Doug Self's "Small Signal Audio Design" and one of the early chapters covers distortion caused by electrolytic capacitors. Skimping on the output capacitors will cause more distortion although whether this is audible depends on how critically you will be listening. It probably isn't important in an mp3 player but for a synth that will be heard through studio monitor grade headphones - particularly a bass synth - it could make a difference.

[BJG145]

OK! Cheers.

I'm currently trying to understand the MIDI input connections, which are based on the Daisy Pod's Type A MIDI TRS input. There are connections labelled "MIDI IN A" and "MIDI IN B" on the header...not sure what the difference is though.

Daisy Bud header:



Daisy Bud schematic:



Daisy Pod schematic:



TRS MIDI guide

What I'm trying to do is send MIDI data to it from a Teensy 4.1.



https://www.pjrc.com/teensy/td_libs_MIDI.html

So it seems I need 3.3V and TX out from the Teensy 4.1 through 47R resistors, and I'm trying to work out what pins to connect them to on the Daisy Bud...

[Folderol]

If you get them swapped, it won't do any harm. It just won't work!

[BJG145]

This is a close-up from the Daisy Bud schematic.



It's not just that I don't know which way round to connect them. I don't understand how to connect them at all.

There's "MIDI IN A" and "MIDI IN B" on the pin header, but I don't know what these represent. The Teensy circuit output has three connections:

Data
+3.3V
Ground

Trying to figure out where they go...

[Folderol]

MIDI A IN is the +3.3V
MIDI B IN is the Data
The plug shell should be connected to the local Ground, and not carried through the cable.

Actually, if the incoming source is 3.3V then R10 should be 47 Ohms - 220 Ohms is for a 5V supply.

[BJG145]

Awesome, thanks!

[James Perrett]

I'd add that the Daisy Pod schematic doesn't make any sense to me.

[BJG145]

Oh good, glad it's not just me then.

[BJG145]

MIDI A IN is the +3.3V
MIDI B IN is the Data
The plug shell should be connected to the local Ground, and not carried through the cable

I've been delving into MIDI circuits to understand why Folderol was right.

When the data signal on MIDI_IN_B is high, it matches the 3.3V at MIDI_IN_A, and no current flows.

When the data signal on MIDI_IN_B is low, current flows from MIDI_IN_A to MIDI_IN_B through the LED in the H11L1V optocoupler.

The D2 diode protects the optocoupler from reverse current if the input polarity gets swapped by mistake, hence:

If you get them swapped, it won't do any harm. It just won't work.

The optocoupler allows signals to pass from the MIDI output device on the left to the MIDI input device on the right, while isolating the two circuits for protection. They're not connected electrically at all, which is why the cable doesn't carry a ground signal.

On reflection, this elaborate connection seems unnecessary just to send MIDI data from Teensy 4.1 to Daisy Seed, and I guess I could probably just connect their data in/out pins directly instead, if I could figure out the code. But it's been interesting and useful to see how it works, and this kind of setup would be needed for hooking the instrument up to other MIDI devices.