Group Buy - Digital Attenuator (Kevin Gilmore)

[audiostar]

10 hours ago, Pars said:

I'm building a couple of the UGS Muse preamps from the french site homecinema.fr / diyaudio. The only part done on both of the is the processor boards, and yes, it was easy to flash. These do use the 72323 chips (which I haven't bought yet).

Hey Pars, the BOM for that preamp says explicitly MUSES72320, are you sure you want to buy the 72323? 
The 72323 is the newer chip, has better specs, need less external parts and is cheaper. Unfortunately not compatible with the the older 72320.

We should be able to take out the volume control portion of the code from there. I am making a balanced board atm that will sit at the amp's input and connect via cable to the controller board holding the encoder and the microprocessor itself. But the STM32G031K8T chip is sold out everywhere; ebay and AE remain only.

Edited April 15, 2022 by audiostar

[audiostar]

Here the actual attenuator based on the 72323 for one balanced channel (the 72323 already includes two channels, so + and - for us). The three lines on the left are coming from the controller.  The jumper J1 sets the chip addressing for the left/right channel. Not sure if we need the 100R resistors on the output with Kevin amps, as well an opamp (like the Muses NJM2082), which according to the tech specs is recommended. 

Edited April 14, 2022 by audiostar

[audiostar]

Here the specs for the 72323, page 10 and up is about the programming. 

[audiostar]

Howdy,

I was wondering how are people using (and controlling) Kevin's latest digital attenuator? Does everybody has its own solution, i.e. control board? ATmega328p based, something else with own software, using standard Arduino with wires hooked to the encoder and attenuator (like JoaMat was showing)? 

I don't see the GB included anything but the attenuator board and parts for it. Thanks much for any enlightening 🙂 

Edited April 19, 2022 by audiostar

[kevin gilmore]

the original larger boards were compatible with amb's lcduino with no software changes.

the smaller board i did with the color touch screen programming thing from australia.

[migo]

Hi @Kerry! I'm tinkering with rev 2.1 of digital attenuator board here and last available controller code from 2017. I'm using 100k pot here, and readings from pot are very strange. At full travel it only reads about 3.3V and is reading only last 10% of pot travel. When you go below 90% of travel analogread returns 0. Pot is OK, I've tested it another sketch with bare code for pot readings and it reads just fine whole pot travel:

Analog Reading: 0, Voltage =0.00v, Resistance=0.00 ohms, Percentage: 0%
Analog Reading: 1, Voltage =0.00v, Resistance=97.75 ohms, Percentage: 0%
Analog Reading: 13, Voltage =0.06v, Resistance=1270.77 ohms, Percentage: 1%
Analog Reading: 56, Voltage =0.27v, Resistance=5474.10 ohms, Percentage: 5%
Analog Reading: 194, Voltage =0.95v, Resistance=18963.83 ohms, Percentage: 18%
Analog Reading: 295, Voltage =1.44v, Resistance=28836.76 ohms, Percentage: 28%
Analog Reading: 624, Voltage =3.05v, Resistance=60997.07 ohms, Percentage: 60%
Analog Reading: 775, Voltage =3.79v, Resistance=75757.58 ohms, Percentage: 75%
Analog Reading: 979, Voltage =4.78v, Resistance=95698.93 ohms, Percentage: 95%
Analog Reading: 1019, Voltage =4.98v, Resistance=99609.00 ohms, Percentage: 99%
Analog Reading: 1023, Voltage =5.00v, Resistance=100000.00 ohms, Percentage: 100%
Analog Reading: 1023, Voltage =5.00v, Resistance=100000.00 ohms, Percentage: 100%

Can you post updated code please? Or give me a hint where to look for this catch?

Thank you very much!

Kind regards,

Milan

 

[Kerry]

Hi Milan,

You should be using a 10k linear pot for the controller. 

That might be older code, but it’s been a while and it will take some time for me to find and verify a newer version. 

Best,

Kerry

[audiostar]

Hi Kerry,

this message here shows a 100k pot used in conjunction with the v2 board, the Bourns RK16812MG099. Here as well. 

The linear taper 10K pot was used in older v1 dual-board attenuator, that is compatible with LCDuino. 

migo seems to be using a v2 board, so the 100k pot looks to be about right, no?

Edited August 22, 2022 by audiostar

[Kerry]

At some point I switched to a 10K pot.  I liked the characteristics better with the 10K pot.  My newest boards use 10K pot as well.

[audiostar]

3 hours ago, Kerry said:

At some point I switched to a 10K pot.  I liked the characteristics better with the 10K pot.  My newest boards use 10K pot as well.

Doesn't this require some changes in the software? Cause the posted software was solely developed for the 100k pot, correct me if I'm wrong.

[migo]

13 hours ago, Kerry said:

Hi Milan,

You should be using a 10k linear pot for the controller. 

That might be older code, but it’s been a while and it will take some time for me to find and verify a newer version. 

Best,

Kerry

Hi @Kerry, thank you. I've 10k pot here too, but it is the same, maybe a bit worse because max. value from analogread I can get is <600. I've changed pot type in controller SW to 10k.

It would be great to have more recent version of controller SW available. Hopefully you will find some time to help here I would appreciate it a lot.

Thank you!

Kind regards,

Milan

 

[migo]

8 hours ago, audiostar said:

Doesn't this require some changes in the software? Cause the posted software was solely developed for the 100k pot, correct me if I'm wrong.

There are three pot values supported in currently available controller SW from 2017: 10k, 50k and 100k. You can chose one used in your case by changing one variable.

[migo]

21 hours ago, Kerry said:

At some point I switched to a 10K pot.  I liked the characteristics better with the 10K pot.  My newest boards use 10K pot as well.

Hi @Kerry, I've sorted out problems with incorrect pot reading now.

Relays are clicking at every position change but output resistance doesn't always change. I'm not sure if measured resistance is OK. At position 127 ohmmeter reads 10.2k, at position 126 ohmmeter reads 9.58k, at 0 position is reads 8.3R, at position 1 to 31 it reads 245.8R, at position 32 it reads 246.8, position 63 is 251R, then at position 64 it is 1.409k etc.

Attenuator PCB used is V2.1.

What is the best way of troubleshooting this?

Thank you! Kind regards,

Milan

 

 

[kevin gilmore]

put a 1.5v battery on the input, and cycle thru the steps and watch the output voltage.

[migo]

Thank you Mr. Gilmore for help and all projects you have published for us to play with. I've wired 3.3V from arduino to one input and measured output voltage as suggested. It seem to work properly to me:

Looks like Log pot to me like it should do.

Kind regards,

Milan

 

 

 

[audiostar]

Working on a PCB for the pot for controlling Kevin's v2 digital attenuator. It will hold an Arduino Nano v3 piggy bag'ed on to it. Thanks to @migofor testing everything out.

Edited September 1, 2022 by audiostar

[audiostar]

On 10/31/2016 at 2:29 AM, Kerry said:

Also, I added a 20K pull-up resistor on the pot's wiper for the larger board since the it's a 100K/3B taper.  I've done some mapping work in code to give a smooth/consistent volume across the full dial (modeled after my encoder).

Hi Kerry, how is that 20k resistor wired exactly, from the pots viper to Vcc (5V)? Thanks.

Edited September 3, 2022 by audiostar

[Kerry]

I don’t use that anymore. Just the 10K pot. Works much better. 

[audiostar]

1 hour ago, Kerry said:

I don’t use that anymore. Just the 10K pot. Works much better. 

Marvelous, thanks Kerry! 

Could you post the p/n of the 10k pot or is it the same as the 100k but 10k?
Looks like it is the RK16812MG09T. 10k and same 3B tapper and dual unit as the RK16812MG099.

Edited September 3, 2022 by audiostar

[mrjayviper]

can someone please tell me the power requirements (transformer?) for a balanced version of this relay attenuator?

thanks

[migo]

On 9/3/2022 at 11:11 PM, Kerry said:

I don’t use that anymore. Just the 10K pot. Works much better. 

@Kerry Original code you published had 128 volume steps, but we have 8 relays here, with 0.5dB step. Was your intention to mask lower 128 steps to have not too much useless (quiet) steps to turn with the pot? With this is max. attenuation of -127.5dB (volume 0) and next step will be -63dB for volume 1.

Thank you in advance for clarification.

Edited September 18 by migo

[Kerry]

Yes.  At the time, that seemed to be the easiest way to get a reasonable taper curve.

These days for the work we're doing with HeadAmp, we've mapped several curves against the full 256 step capability of the attenuator.

[migo]

Thank you @Kerry! I'm curious, sorry, new mapping approach is based upon fixed table mapping or some smart math?

[Kerry]

There are two mappings. One for the potentiometer, which determines the rotational position of the volume knob, and then a second for the log taper of the audio.