That1580 mini mixer
- mediatechnology
- Posts: 5744
- Joined: Sat Aug 11, 2007 2:34 pm
- Location: Oak Cliff, Texas
- Contact:
Re: That1580 mini mixer
It also occurred to me that you might consider combining the 124X common mode stage with the modulator driver.
Use a "Birt" topology similar to those shown in THAT's DN-140 (figure 11 basic Birt) and the THAT 5263 data sheet figure 12.
http://www.thatcorp.com/datashts/THAT_5 ... asheet.pdf
I did the development work on both of those circuits and the modulation driver works quite well.
It not only saves you a stage it allows the common mode rejection circuit to have lower resistor values than the 124X.
Use a "Birt" topology similar to those shown in THAT's DN-140 (figure 11 basic Birt) and the THAT 5263 data sheet figure 12.
http://www.thatcorp.com/datashts/THAT_5 ... asheet.pdf
I did the development work on both of those circuits and the modulation driver works quite well.
It not only saves you a stage it allows the common mode rejection circuit to have lower resistor values than the 124X.
Woke is dead.
https://ka-electronics.com
https://ka-electronics.com
Re: That1580 mini mixer
can't do the polarity digital because signal is still analog at that point.
the adc happens after the virtual earth summing:
5x2 analog mixer>> stereo analog buffer for adc >>stereo adc>>mcu>>microsd
i thought about what you're saying (sorta) with nixing the 1240's.
pro: save a stage, less noise probably, signal remains differential right through to the adc
con: time, money, redesign. summing will have to be differential.
i already have the boards done and in hand. one channel is populated but not tested.
i think it comes down to noise/distortion. if i won't hear any difference then i'm inclined to leave things as they are. but, convince me!
1240 was not in stock last time i checked mouser and it looks like ina134 is drop in replacement so i might go with those.
not sure how the birt configuration you posted compares to what cirrus recommends.
see figure 2
https://statics.cirrus.com/pubs/appNote/an241-1.pdf
the adc happens after the virtual earth summing:
5x2 analog mixer>> stereo analog buffer for adc >>stereo adc>>mcu>>microsd
i thought about what you're saying (sorta) with nixing the 1240's.
pro: save a stage, less noise probably, signal remains differential right through to the adc
con: time, money, redesign. summing will have to be differential.
i already have the boards done and in hand. one channel is populated but not tested.
i think it comes down to noise/distortion. if i won't hear any difference then i'm inclined to leave things as they are. but, convince me!
1240 was not in stock last time i checked mouser and it looks like ina134 is drop in replacement so i might go with those.
not sure how the birt configuration you posted compares to what cirrus recommends.
see figure 2
https://statics.cirrus.com/pubs/appNote/an241-1.pdf
- Attachments
-
- ADC.pdf
- (66.8 KiB) Downloaded 1824 times
-
- 1580card.pdf
- (106.38 KiB) Downloaded 1821 times
Last edited by weroflu on Sun Feb 27, 2022 10:00 am, edited 1 time in total.
- mediatechnology
- Posts: 5744
- Joined: Sat Aug 11, 2007 2:34 pm
- Location: Oak Cliff, Texas
- Contact:
Re: That1580 mini mixer
For some reason I was thinking each pair of channels fed an A/D.
Forgot you were analog summing.
For the single modulator driver the Birt might still be advantageous.
The Cirrus circuit requires anti-polarity inputs.
The Birt can be fed single-ended or differential but still provides differential output.
One of the advantages of the Birt output.mod driver is that the second inverting op amp stage's noise appears in common mode and is thus rejected by the A/D's CMRR.
If you think you're going to need INA134's I would order them now.
Deliveries on everything is getting crazy long.
Forgot you were analog summing.
For the single modulator driver the Birt might still be advantageous.
The Cirrus circuit requires anti-polarity inputs.
The Birt can be fed single-ended or differential but still provides differential output.
One of the advantages of the Birt output.mod driver is that the second inverting op amp stage's noise appears in common mode and is thus rejected by the A/D's CMRR.
If you think you're going to need INA134's I would order them now.
Deliveries on everything is getting crazy long.
Woke is dead.
https://ka-electronics.com
https://ka-electronics.com
Re: That1580 mini mixer
I got as far as testing audio output from the master bus and testing simple audio recording from both a test tone fed to the master out bypassing the mixer section, and also the regular bus.
All fails.
I didn't have any test point on the preamp cards so I couldn't even probe signals there.
My main suspect was the polarity switches, the footprint is so small that it's impossible to know what's going on.
In any case I'm using this opportunity to do things more properly.
I'm ditching the analog summing and analog polarity switching, and attempting to make this a 5 channel recorder, summing / polarity itb.
The major changes now are going with a differential input adc driver. I looked at Birt, the Cirrus recommended driver,TI driver, and AK. AK is simplest topology (appeals to me), Cirrus I have some questions about, TI uses a FDA - have a few questions about that, and I modelled the Birt circuit a few days ago trying to understand the gain.
I think if Wayne can help out I can just go with the Birt and not try to waste time with the other approaches.
Found another design error/ question. My power rails are +-5v for the That chips and the Cirrus CS5366. Reading the datasheet more carefully, these will swing to +-.3v beyond the rails. So my 5V supply won't cut it. I either run everything at 6.5v-7v or add in a boost dc-dc somewhere for the driver. Do I even want to push that limit and feed the adc right up to those +-.3v levels or just keep the adc at 5V peak to peak.
Questions on the Birt driver as shown in the 5263 datasheet figure 12...
First, what is the proper way to define full scale? Up till now I was thinking with 5v rails, the That1580 sheet says it swings to +-4v. So everything following I was thinking max swing is 8v p/p.
Do I need ac coupling - C11 and C12, or does the servo on the 5171 eliminate enough dc at the output of the 1580? Figure 12 does not show any caps.
https://statics.cirrus.com/pubs/appNote/an241-1.pdf
I don't know how relevant Cirrus's buffer recommendations are, or whether I should try to incorporate them into the Birt circuit.
- they use a high pass filter with the ac coupling cap.
- I have also seen people add a low pass with corner ~220khz
I think C17 in the 5263 datasheet is a miller compensation cap? Over my paygrade. Should I stay with the 2n2 value or is there bench experimentation needed.
R12 changes to 91r as per the Cirrus circuit.
Gain is set with R11/R9 (and R13/R10)? I thought that R12 and R15 are in the FB loop but running the sim showed just R11/R9 set gain.
All fails.
I didn't have any test point on the preamp cards so I couldn't even probe signals there.
My main suspect was the polarity switches, the footprint is so small that it's impossible to know what's going on.
In any case I'm using this opportunity to do things more properly.
I'm ditching the analog summing and analog polarity switching, and attempting to make this a 5 channel recorder, summing / polarity itb.
The major changes now are going with a differential input adc driver. I looked at Birt, the Cirrus recommended driver,TI driver, and AK. AK is simplest topology (appeals to me), Cirrus I have some questions about, TI uses a FDA - have a few questions about that, and I modelled the Birt circuit a few days ago trying to understand the gain.
I think if Wayne can help out I can just go with the Birt and not try to waste time with the other approaches.
Found another design error/ question. My power rails are +-5v for the That chips and the Cirrus CS5366. Reading the datasheet more carefully, these will swing to +-.3v beyond the rails. So my 5V supply won't cut it. I either run everything at 6.5v-7v or add in a boost dc-dc somewhere for the driver. Do I even want to push that limit and feed the adc right up to those +-.3v levels or just keep the adc at 5V peak to peak.
Questions on the Birt driver as shown in the 5263 datasheet figure 12...
First, what is the proper way to define full scale? Up till now I was thinking with 5v rails, the That1580 sheet says it swings to +-4v. So everything following I was thinking max swing is 8v p/p.
Do I need ac coupling - C11 and C12, or does the servo on the 5171 eliminate enough dc at the output of the 1580? Figure 12 does not show any caps.
https://statics.cirrus.com/pubs/appNote/an241-1.pdf
I don't know how relevant Cirrus's buffer recommendations are, or whether I should try to incorporate them into the Birt circuit.
- they use a high pass filter with the ac coupling cap.
- I have also seen people add a low pass with corner ~220khz
I think C17 in the 5263 datasheet is a miller compensation cap? Over my paygrade. Should I stay with the 2n2 value or is there bench experimentation needed.
R12 changes to 91r as per the Cirrus circuit.
Gain is set with R11/R9 (and R13/R10)? I thought that R12 and R15 are in the FB loop but running the sim showed just R11/R9 set gain.
- mediatechnology
- Posts: 5744
- Joined: Sat Aug 11, 2007 2:34 pm
- Location: Oak Cliff, Texas
- Contact:
Re: That1580 mini mixer
To set the amount of attenuation in the Birt modulator driver I would work backwards from the 2V RMS FS level of the CS5366.
The CS5366 modulator driver I reference below is unity gain.
Using the Birt you could equalize the clipping points of the mic preamp and A/D to determine the proper amount of attenuation.
That's one approach.
Equalizing the clipping points requires a lot of gain in the front end to be thrown away to prevent overloading the 2V RMS A/D.
THAT have a paper which addresses the optimum preamp gain I need to find for you.
To answer your questions about the Birt vs Cirrus drivers...
I think you'd be better off using figure 15 here: https://www.micro-semiconductor.com/dat ... 6-DQZR.pdf
It's unity gain which avoids having to add a lot of gain in the preamp only to attenuate it later to drive the A/D.
The input is AC-coupled into a relatively high impedance (10K) compared to the Birt so the caps are smaller.
If you use this approach common mode rejection is realized in the CS5366; the Birt has CM rejection inherently but is limited by resistor match.
For the values of the "in the loop C", resistor values and modulator cap I'd stick with Cirrus' recommended values.
The CS5366 modulator driver I reference below is unity gain.
Using the Birt you could equalize the clipping points of the mic preamp and A/D to determine the proper amount of attenuation.
That's one approach.
Equalizing the clipping points requires a lot of gain in the front end to be thrown away to prevent overloading the 2V RMS A/D.
THAT have a paper which addresses the optimum preamp gain I need to find for you.
To answer your questions about the Birt vs Cirrus drivers...
I think you'd be better off using figure 15 here: https://www.micro-semiconductor.com/dat ... 6-DQZR.pdf
It's unity gain which avoids having to add a lot of gain in the preamp only to attenuate it later to drive the A/D.
The input is AC-coupled into a relatively high impedance (10K) compared to the Birt so the caps are smaller.
If you use this approach common mode rejection is realized in the CS5366; the Birt has CM rejection inherently but is limited by resistor match.
For the values of the "in the loop C", resistor values and modulator cap I'd stick with Cirrus' recommended values.
Woke is dead.
https://ka-electronics.com
https://ka-electronics.com
- mediatechnology
- Posts: 5744
- Joined: Sat Aug 11, 2007 2:34 pm
- Location: Oak Cliff, Texas
- Contact:
Re: That1580 mini mixer
Here is the THAT paper I was thinking about: http://www.thatcorp.com/datashts/Interf ... erters.pdf
My advice goes somewhat contrary to "gain early attenuation late" because your situation is a little different.
You have size and power dissipation concerns.
Trying to remember: Is there an input pad or line input?
WRT the modulator driver.
It seems to me that if the output of the mic preamp were direct-coupled to a low R value Birt circuit that a lot of power would be wasted burning off the preamp's DC CM voltage.
The resistor values are small due to noise concerns.
If AC-coupled into the Birt modulator driver the coupling caps would have to be pretty large and take up a lot of space.
The Cirrus driver has a bias resistor on the non-inverting input that can be made high reducing the capacitor size.
No CM voltage from the preamp is burned off so there's no wasted power.
My advice goes somewhat contrary to "gain early attenuation late" because your situation is a little different.
You have size and power dissipation concerns.
Trying to remember: Is there an input pad or line input?
WRT the modulator driver.
It seems to me that if the output of the mic preamp were direct-coupled to a low R value Birt circuit that a lot of power would be wasted burning off the preamp's DC CM voltage.
The resistor values are small due to noise concerns.
If AC-coupled into the Birt modulator driver the coupling caps would have to be pretty large and take up a lot of space.
The Cirrus driver has a bias resistor on the non-inverting input that can be made high reducing the capacitor size.
No CM voltage from the preamp is burned off so there's no wasted power.
Woke is dead.
https://ka-electronics.com
https://ka-electronics.com
Re: That1580 mini mixer
I've been through so many revisions that a lot of the info from the older posts isn't so relevant.
As of this revision my power supply is:
9v switching wall wart-->
lt3045 regulator for +7v, +3.3v
inverting converter 9v>-8v>lt3094>-7v
boost converter >50v>lt3012 regulator>48v
not much power concerns anymore. initially i was using much higher rails unnecessarily and the ltm8049 power module.
No input pads or line inputs, just mic ins
Attached is a pic of current layout. There's still a bit of room on the channel cards.
yellow caps are polymer not tant
As of this revision my power supply is:
9v switching wall wart-->
lt3045 regulator for +7v, +3.3v
inverting converter 9v>-8v>lt3094>-7v
boost converter >50v>lt3012 regulator>48v
not much power concerns anymore. initially i was using much higher rails unnecessarily and the ltm8049 power module.
No input pads or line inputs, just mic ins
Attached is a pic of current layout. There's still a bit of room on the channel cards.
yellow caps are polymer not tant
- mediatechnology
- Posts: 5744
- Joined: Sat Aug 11, 2007 2:34 pm
- Location: Oak Cliff, Texas
- Contact:
Re: That1580 mini mixer
Rather than a 9V wall-wart have you considered using 5V USB C power?
It's ubiquitous at this point and low-cost cell phone batteries could be used to run it.
WRT DC-coupling the input to the modulator I'm calculating about 3.2 mA of CM DC being burnt off per channel. (2K Rin).
That's 268 mW of wasted power for 6 channels.
With +/-7V supplies is looks like you need about 11 dB attenuation in the A/D modulator driver to equalize the clipping and A/D FS points.
With 14V supplies the preamp should be capable of about +18 or +19 dBu differential output.
Alternatively you could run the preamp at 11 dB lower gain and use the AC-coupled unity gain Cirrus driver.
With the later approach the A/D is providing CM rejection and for large CM signals some analog DR is used.
It's ubiquitous at this point and low-cost cell phone batteries could be used to run it.
WRT DC-coupling the input to the modulator I'm calculating about 3.2 mA of CM DC being burnt off per channel. (2K Rin).
That's 268 mW of wasted power for 6 channels.
With +/-7V supplies is looks like you need about 11 dB attenuation in the A/D modulator driver to equalize the clipping and A/D FS points.
With 14V supplies the preamp should be capable of about +18 or +19 dBu differential output.
Alternatively you could run the preamp at 11 dB lower gain and use the AC-coupled unity gain Cirrus driver.
With the later approach the A/D is providing CM rejection and for large CM signals some analog DR is used.
Woke is dead.
https://ka-electronics.com
https://ka-electronics.com
Re: That1580 mini mixer
Re: power supply, yes I had a few different power supply approaches along the way, but I'm very happy you brought this up now. I got a bit lazy for the last p/s rev. using +-5v rails. I was thinking 6v wall wart input would not be too wasteful if I only had to drop 1V for the +5v. Also 3.3V now needs much more current than earlier versions via the teensy, sdio card interface, so I think I need to go back to sepic configuration on the +7 and 3.3v rails, this will save a lot of waste.
A low profile battery source was of course in my mind too but for development the wall wart is very easy. Going back to sepic dc-dc conversion will enable 5v input again.
Thanks for all the info on the buffer. I guess it never occurred to me to just turn the gain down. Maybe that approach plus some sort of clamping on the adc differential inputs? Still have to think on this a bit. I read the That powerpoint and I'm inclined to go with the gain early/ attenuate later rule.
Just out of curiosity there is no audio bandpass filtering on the That/Birt buffer,how important do you think it is?
A low profile battery source was of course in my mind too but for development the wall wart is very easy. Going back to sepic dc-dc conversion will enable 5v input again.
Thanks for all the info on the buffer. I guess it never occurred to me to just turn the gain down. Maybe that approach plus some sort of clamping on the adc differential inputs? Still have to think on this a bit. I read the That powerpoint and I'm inclined to go with the gain early/ attenuate later rule.
Just out of curiosity there is no audio bandpass filtering on the That/Birt buffer,how important do you think it is?
- mediatechnology
- Posts: 5744
- Joined: Sat Aug 11, 2007 2:34 pm
- Location: Oak Cliff, Texas
- Contact:
Re: That1580 mini mixer
While I'm bringing things up have you considered the THAT626X?
http://www.thatcorp.com/datashts/THAT_6 ... asheet.pdf
The Birt does have rolloff owing to the large value feedback caps.
I need to dig into my notes to remember how we set the C-values to control the "hump" in the output.
http://www.thatcorp.com/datashts/THAT_6 ... asheet.pdf
The Birt does have rolloff owing to the large value feedback caps.
I need to dig into my notes to remember how we set the C-values to control the "hump" in the output.
Woke is dead.
https://ka-electronics.com
https://ka-electronics.com