Multichannel Audio DSP Field Mixer Recorder

Aluminium emi between digi and analogue board

Making the field recorder into a small square box 15x15x7cm

The plan here is to mount the DSP digi card and analouge main board to

a 2mm 10x10cm sheet of aluminium which will hopefully shield the EMI from

digi board to analogue board ( preamps )

Perspex demo case with mini xlr IO and 2 x headphone 1/4 inch jacks

ok finally have enough bits to get the new analogue board running without a box for it !

I'm tempted to just buy some aluminium sheet and just make the prototype box ! I watched some interesting videos of people using brazing rods to weld aluminium with a gas torch. If not aluminium maybe just some acrylic sheet and epoxy glue 

1/ The headphone sockets should be on the underside of the the connector board to allow more clearance with the miniXLR sockets and to fit into the corner of the box better.

2/ The ADC - DAC card should be flush with the top of analogue and Digi boards allowing the power and SD cards to be flush with the edge of the box. Also digi board is not level with analogue board!

3/ power back plane could do with shifting all the pwr cards left 2 cm's

4/ the SPI lines for preamp gain control will be 16cm from digi board ! a bit far !!

5/ I should put the V sense dividers and reference on the power backplane. Just thought of that as I type this !

6/ dump OLED screen and use that area for 12v to 3.3v 1amp DC-DC rather than steeling 5V power from the analouge pwr board.

7/ where to put the isolated 9v - 18v 12v regulator and do I include a charger for maybe 4 or 8 x 18650 3.7v Battery's ?

here's a pic of the new analogue board.

It has optical isolated power switching for preamp channels

Also has +/-5 and +-10v power monitoring with reference 3.3v using

resistor dividers.

It all looks bigger than it will be when its soldered together without

headers as connectors !

the power backplane on the right will probebly get smaller.

It will have 8 analogue differential channels in and 4 out.

2 high quality stereo headphone drivers

and timecode IO.

The ADC DAC board at the back will bridge over directly to the digital

board.

new backplane nearly finished ! 8 preamp cards fit into a smaller space.

I am using a shocking amount of SSR's ( solid state relays ) 2 for each preamp channel.

8 on top and 8 on the bottom of the board. Yuk !! Why ? because I there cheap and I know they work.

I will run wires direct to the micro controller breakout connectors on my main mixer board so I don't

need to go to 4 layer board. Its a prototype and the price difference is large.

The max 4mA required for each optoMos SSR pair ( +/- 5v ) can deliver 120mA per channel which

allows plenty of headroom to fire up the PGA2500's 

The ADC_DAC card will plug directly onto the analouge board and bridge to the main Digi board that has the ARM and DSP.

I have a separate connector card and power card that will supply the analogue power.

Also new and really exciting is the OVR connector. This will send over level signals from the individual cards 

to the micro that will be the hardware limiter. The micro can trim back the gain in an instant. hopefully within 20ms or so.

Lastly rail Voltage sense setup with divider resistors and a reference 3.3v. This will be so I can monitor the voltages

of both the +/-5v and +/-10v or whatever I set the higher rails to for Headphone and driver cards.

Theoretically I can shutoff all the preamp cards if there is to much difference in the +/-5v rails as protection !!

Eventually I'd like to experiment with making a tracking dual rail supply MC controlled !!

I'll have to see how accurate my voltage divider setup is 1st !

I found much smaller smd SSR's by Toshiba but at 2$ a go that makes 32$ just to switch the preamps on/off with

isolation. mmmm a bit to pricey for a prototype :)

excited about my new smaller thinner preamp cards using 1.27mm headers.
so far they seam to work as before. I kept some space by the optical relay as I expect there to be more front end circuitry. I hope to find a smaller 1 form c opto relay to use. found a toshiba that looks promising that's 1/4 of the size. The 60v 1Amp clamping protection diodes have started to loo very large so working on finding some smaller diodes to.  the gold caps to the right are DC blocking caps before the ADC. The PGA2500's output some DC which needs filtering out.

The dual rail cards are quite long at 80mm but output +/- 1 amp at incredible low noise using lm43602's and TI ultra low noise regulators.
I have 1 card for +/-5v and one for +/-15v ( or I might use +/- 9v )  depending
on how much power I want on the master outputs of the line drivers.
I should change the conductors for +/- 15v setup from 6.8uH to 15uH.
The negative rails needs a lower in series resistance inductor 150megOhms. the + rail works with 75megOhms.
both cards are set to  run at 1mhz switch frequency but are not running in sync yet.
I'm interested in knowing how you can use an ADC on say the ARM micro to monitor the DCV of each power channel and then even using a digipot fine tune the accuracy of each rail. fine tuning the Voltage is perhaps overkill but monitoring the rails could be good way to self test the power is healthy before potentially frying amp chips in case a power card went down. maybe i'd need to use a higher voltage capable adc to feed the results to the micro ? and how about the negative voltages? would I need an ADC capable of neg voltage ? or use a voltage buffer for each channel and invert the neg rail?

Although I was fairly happy with the shrinking down of the 1st cards ( still more to do ) the 3.3v 2amp power card was a complete fail!!!

1st I messed up the voltage divider traces. I tried adding the correct resistor to ground to set the TI tpsm84209 output voltage to 3.3v

Without the regulator IC in circuit the output fluctuates all over the place.

I'm thinking so far using the through hole resistor made the voltage divider loop to large. ( the datasheet warns against making that loop large ! )

Although the caps are within voltage spec I'm not sure about the in series resistance. This is were to be honest I'm entering a new area. I don't have a LCR meter and I don't even know if there is LCR mode on my keysight 34461A bench meter. ( better check the manual but dont think so ) 

maybe if I generate 100khz square wave with the mixer recorder I could get the in series resistance of the capacitors I used that way ( the old mixer recorder prototype is still working ) . never used an LCR , no idea yet how it works. looks like I'll need one eventually.

Even the picture I took of this little 1 x 1 cm board failed. it came out upside down and blurred. it cost 1$ , so I think the best thing is to make a better effort with the pcb design and start from there!

These 4 cards will slot into a backplane to make

+/-15v(+/-1A) line drivers and headphone

+/-5v(+/-1A) preamps and adc + dacs

48v(135mA) phantom power

3.3v(2A) digital power

I just finished soldering up a smaller modular phantom power board.

I did look into making the 12v -> 48v boost converter side without IC's ( discrete components ) but it would have been to large for what I need. I tested on a dummy load and can produce 134mA 47.5V without the board warming up over long periods.

I've started to use thinner circuit boards now to to reduce weight. I should have used 1.27mm headers for this board. I went against it thinking its a power board but at maximum 134mA a smaller header would have been fine.

This is just the start of shrinking all my proto boards down !

Next is the dual +/-5v and +/-15v boards and new smaller preamp boards with end cutouts for the DC blocking caps :)

DMA is working from SAI interfaces -> memory & memory -> uSD cards for recording and playback.

I have UART / bluetooth RX TX working over DMA to.

I'm not using DMA for SPI or both I2C1 and I2C2 buses.

I might need to for the I2C readback from DSP if I start reading a lot of meters and LTC timecode etc.

I'm thinking I will make one large packet from all meter data. its 1 byte per meter with a few bytes for header info for each meter type. I think I can get all the useful data back to the android app in approx 45bytes per packet streaming at about 1kbps. 

The main problem seams to be the app inventor app can't keep up with the speed of the packets being sent from the field recorder. I did slow down the TX rate so it could also be the way the individual screens are cleared.

I have to say a massive thanks to ABG on the MIT app inventor forum for helping me figure out a fast way to receive the 10 byte packets. My caveman code in app inventor was just not working quick enough :)

https://groups.google.com/forum/?utm_medium=email&utm_source=footer#!msg/mitappinventortest/C2KYLiD1k0g/w_nyJ6OLAgAJ

I thought my adaption of ABG 's example code was working at the end of the thread but I was wrong. I had to slow down the packet stream from the ARM since it was just way to fast.

I'll hopefully be able to clean up the glitches and move on from android app designer. Not my thing !!! If the gain peek metering works I will add metering to main faders and mix faders. There will also be a screen with all faders working at once!! if its fast enough !

I think I need to switch my BT dongle to BTLE 4. I read its better at streaming data.