This seems somewhat following your thinking Hans. I did some loose matching of the Zetex devices specific only to matching the emitter voltage differences being selected as in the test fixture below. This is to the keep potentials ultimately appearing across the coils low, whereupon once in place this would drop. Subsequently it is necessary to add resistance across Rx1 or Rx2 in the collectors to trim the output to zero Volts.
Before simulations I wasn't understood the extent of the difference in the CMRR of variances to the positive and negative power supplies, hence any adjustability on the emitter side warrants avoidance. From the simulations there is about 60db of CMRR for 3kOhm values changed by 0.1 %. 60dB seems a reasonable target in order to keep the -ve regulators simplistic.
As it turns out, using two Zetex devices matched to about 2mV difference required a 100k Ohm resistor to be placed across one of the collector resistors for the output balance to be within the working range of the output servo's. With the Zetex's glued together and the servo set free the movements held within the boundaries of control, though only marginally trustworthy for high gain (exceeding 1 V RMS per side). In contrast, the matched pair SSM device didn't require any trimming of the 3 kOhm values at all (the servo was operating at less than a 1 volt variance at the outputs). It so happened that when I removed the trimmer in the negative side it worked 4x better (badly adjusted).
Before simulations I wasn't understood the extent of the difference in the CMRR of variances to the positive and negative power supplies, hence any adjustability on the emitter side warrants avoidance. From the simulations there is about 60db of CMRR for 3kOhm values changed by 0.1 %. 60dB seems a reasonable target in order to keep the -ve regulators simplistic.
As it turns out, using two Zetex devices matched to about 2mV difference required a 100k Ohm resistor to be placed across one of the collector resistors for the output balance to be within the working range of the output servo's. With the Zetex's glued together and the servo set free the movements held within the boundaries of control, though only marginally trustworthy for high gain (exceeding 1 V RMS per side). In contrast, the matched pair SSM device didn't require any trimming of the 3 kOhm values at all (the servo was operating at less than a 1 volt variance at the outputs). It so happened that when I removed the trimmer in the negative side it worked 4x better (badly adjusted).
Just for fun I simulated a basic MM implementation (since playing around with LTSpice I am now starting to feel "guilty" for not doing other things that are far more important... things I could be playing at instead). A potential problem with this implementation is that it seems could have more serious noise related consequences that I don't know how to investigate yet with LTSpice. Nevertheless it is interesting how simplistically a single ended MM input can convert to balanced differential outputs with DC servo'ing, that if the set point is increased can also serve as a basic form of rumble filter as the so-called servo is embedded in the RIAA, behaving more as an active filter inside a passive one.
Had busy week, not much time except few minutes here and there... But I see Im not the only one here with same issue.
@ Hans, I gave another look at amp you showed us a week ago, and one thing is puzzling me there, might be easier to look at second picture with model and simulation:
It looks like there is no cross feedback between first 2 opamps, basically they are 2 amps in parallel, that means there is no any CMR?
Some CMR is with integrator of second stage, but if there is no CMR in first stage, common signals are already amplified and that one has hard job, also completely dependent on extra precise matching of 4 x 100Ohm resistors. Differential outputs still have no CMR......
What did I miss this time?
@ Hans, I gave another look at amp you showed us a week ago, and one thing is puzzling me there, might be easier to look at second picture with model and simulation:
It looks like there is no cross feedback between first 2 opamps, basically they are 2 amps in parallel, that means there is no any CMR?
Some CMR is with integrator of second stage, but if there is no CMR in first stage, common signals are already amplified and that one has hard job, also completely dependent on extra precise matching of 4 x 100Ohm resistors. Differential outputs still have no CMR......
What did I miss this time?
Hi Dražen,
Sorry, but could you please refer to the relevant posting #nr.
Of all the images that passed, I’m having a bit of a trouble to understand which one you mean.
Hans
Sorry, but could you please refer to the relevant posting #nr.
Of all the images that passed, I’m having a bit of a trouble to understand which one you mean.
Hans
Hi Hans,
Sure, post #660 this one
https://www.diyaudio.com/community/attachments/noise-current-amp-jpg.1377054/
And this one
https://www.diyaudio.com/community/attachments/current-amp-jpg.1377
Sure, post #660 this one
https://www.diyaudio.com/community/attachments/noise-current-amp-jpg.1377054/
And this one
https://www.diyaudio.com/community/attachments/current-amp-jpg.1377
Of course there is cross feedback through the feedback resistors from minus input to opamp output and from both minus inputs through the Cart.
CMRR is very good because with CM signals, no current is flowing through the Cart an thus no current through both feedback resistors.
In the Diff to SE converter, with the four 100R resistors to be within 0.1R, CMRR will be below 60dB, but I used 0.01% resistors.
Hans
Oh jeee, I missed to observe that source, chart, is low resistance and part of the feedback, that was the missing point, makes complete sense...
That does it.
One contemplation about using cartridge R as being Rg, or gain setting resistor. This kind of schematics is being used on this thread a lot.
Now I have Benz Micro out of the arm so I could measure its R more accurately. Also measured Sumiko Blue point. That's what I have from better cartridges.
Measurement is by standard Fluke, I contact sprayed and cleaned, used same cable arrangements for 2 channel R measurements, the difference is what counts, not actual values:
Sumiko Blue point:
Channel 1 : 130.8 Ohm
Channel 2: 128.9 Ohm
That's 1.45% difference
Benz Ruby 2 (almost 10x more expensive than Sumiko):
Channel 1 : 40.8 Ohm
Channel 2: 37.7 Ohm
That's 7.6% difference
I went to google to seek for this subject, seems everyone that was looking at this found similar differences.
When we set the gain with cartridge R, obviously we will get 2 different gains for 2 different channels. Doesn't look good 🤔
Funny thing is that channel with lower R will most likely have lower V out too, so higher gain is actually appreciated to balance channels,
Any experience on this subject?
Cheers,
Dražen
Now I have Benz Micro out of the arm so I could measure its R more accurately. Also measured Sumiko Blue point. That's what I have from better cartridges.
Measurement is by standard Fluke, I contact sprayed and cleaned, used same cable arrangements for 2 channel R measurements, the difference is what counts, not actual values:
Sumiko Blue point:
Channel 1 : 130.8 Ohm
Channel 2: 128.9 Ohm
That's 1.45% difference
Benz Ruby 2 (almost 10x more expensive than Sumiko):
Channel 1 : 40.8 Ohm
Channel 2: 37.7 Ohm
That's 7.6% difference
I went to google to seek for this subject, seems everyone that was looking at this found similar differences.
When we set the gain with cartridge R, obviously we will get 2 different gains for 2 different channels. Doesn't look good 🤔
Funny thing is that channel with lower R will most likely have lower V out too, so higher gain is actually appreciated to balance channels,
Any experience on this subject?
Cheers,
Dražen
Without a Kelvin (4-wire) connection those resistance readings may still be misleading, an ohm or two difference can be down to multimeter selector switch or probe contact resistance variation - the gold plating wears off eventually.
What matters for pick-up performance is the magnetic flux coupling, not wire resistance, and that's much harder easy to measure - in fact the variation in the permanent magnet strength or geometry may be the dominant factor anyway.
Comparing channels with a test disc is probably most meaningful.
What matters for pick-up performance is the magnetic flux coupling, not wire resistance, and that's much harder easy to measure - in fact the variation in the permanent magnet strength or geometry may be the dominant factor anyway.
Comparing channels with a test disc is probably most meaningful.
Hi Mark, I'm fully aware of this issue. That's why I cleaned contacts and more importantly, did not do any cable, instrument movements between measurements of different channels. In about 10 consecutive measurements got the difference between channels I presented. Ohm up and down is measurement mistake, but I think the difference between channels is rather accurate.
Agree to this too. But what I examine is use of cart R as feedback resistor, as we do in many designs. This is preamplifier question where just Chart R plays role. Magnetic coupling is absolutely important, but off subject.
Hello,
I would like to make one preliminary summary on specialized microphone pream chips. Like SSM2107 that I'm using,but also similar ones like newer SSM2019, THAT151x, some INA-s.
All these chips are specialized to amplify microphone signals delivered by long cable, CMRR will be more than excellent on these due to laser trimmed feedback resistors, noise very low too, THD absolutely acceptable.
One issue they have (and is might be not an issue considering LP noise) is they noise performance is optimized for microphones source resistance , 100 to 300 Ohm, while LO MC cartridges are in level of 5 to 50 Ohm.
Below I made ABC sketches of 3 variations using mic preamps in normal voltage mode. All are just focusing on first stage, RIAA compensation and second stage you can imagine.
Variation A:
Just a single mic amp; about 0.9nVsqrtHz, very good CMR and gain up to 60db.
But consider, if one is using high output MC like I have Sumiko Blue point, its R source is about 130 Ohm, very well in range of microphones. Mic opamp should excel here.
Otherwise, It is way to high R for LOMC, and way to low for MM (typical Rsource 600 to 3000 Ohm), but for high output MC it could be just nice.
Variation B:
True balanced what I used through years with 40 Ohm cart. Mic amps input transistors are effectively in parallel , lowering Rin and noise and allowing 66db gain. IMO there is nothing wrong in using this with highish Rsource MC carts.
Variation C:
We can of course use circuit B and add chips in parallel; further reducing Rin and noise, becoming more adequate for lower Rsource LOMC carts.
Matching Rg resistors now becomes more critical, but manageable.
These circuits will not reach noise levels of trains impedance input with ZTX transistors. However they need only few external components ( one Rg per chip), have great CMRR , guaranteed THD and noise performance, In addition not that expansive at all (guess because the need for mic preamps is much greater than for MC preamps, so it makes sense in industrial production costs).
For high Rsource MC cartridges (HI out MC) they might as well be optimal.
These are just my toughs on mic preamp chips subjects, to wrap up.
Cheers,
Dražen
I would like to make one preliminary summary on specialized microphone pream chips. Like SSM2107 that I'm using,but also similar ones like newer SSM2019, THAT151x, some INA-s.
All these chips are specialized to amplify microphone signals delivered by long cable, CMRR will be more than excellent on these due to laser trimmed feedback resistors, noise very low too, THD absolutely acceptable.
One issue they have (and is might be not an issue considering LP noise) is they noise performance is optimized for microphones source resistance , 100 to 300 Ohm, while LO MC cartridges are in level of 5 to 50 Ohm.
Below I made ABC sketches of 3 variations using mic preamps in normal voltage mode. All are just focusing on first stage, RIAA compensation and second stage you can imagine.
Variation A:
Just a single mic amp; about 0.9nVsqrtHz, very good CMR and gain up to 60db.
But consider, if one is using high output MC like I have Sumiko Blue point, its R source is about 130 Ohm, very well in range of microphones. Mic opamp should excel here.
Otherwise, It is way to high R for LOMC, and way to low for MM (typical Rsource 600 to 3000 Ohm), but for high output MC it could be just nice.
Variation B:
True balanced what I used through years with 40 Ohm cart. Mic amps input transistors are effectively in parallel , lowering Rin and noise and allowing 66db gain. IMO there is nothing wrong in using this with highish Rsource MC carts.
Variation C:
We can of course use circuit B and add chips in parallel; further reducing Rin and noise, becoming more adequate for lower Rsource LOMC carts.
Matching Rg resistors now becomes more critical, but manageable.
These circuits will not reach noise levels of trains impedance input with ZTX transistors. However they need only few external components ( one Rg per chip), have great CMRR , guaranteed THD and noise performance, In addition not that expansive at all (guess because the need for mic preamps is much greater than for MC preamps, so it makes sense in industrial production costs).
For high Rsource MC cartridges (HI out MC) they might as well be optimal.
These are just my toughs on mic preamp chips subjects, to wrap up.
Cheers,
Dražen
Hi Dražen,
Something went wrong in version C where both inputs of the amp on top are short circuited.
About noise: with the ssm2017 you will need a high gain like 60dB to get a low input noise, but when doing that, 0.95nV/rtHz for most MC carts will be good enough even for lesser output MC’s.
S/N for version B improves even by 3dB and at the same time giving a diff output.
So with version B, only a very exotic MC cart wouldn’t meet the minimum required S/N.
Hans
Something went wrong in version C where both inputs of the amp on top are short circuited.
About noise: with the ssm2017 you will need a high gain like 60dB to get a low input noise, but when doing that, 0.95nV/rtHz for most MC carts will be good enough even for lesser output MC’s.
S/N for version B improves even by 3dB and at the same time giving a diff output.
So with version B, only a very exotic MC cart wouldn’t meet the minimum required S/N.
Hans
Dear Hans, you have sharp eye, and I have some issues with concentration... Here is corrected one:
Nevertheless you got the point, even dwg was wrong.
I'm happy we agree on mic amps, IMO for many DIY-ers this is fastest and very plausible way to get to very good results.
For more extreme LOMC carts, I'm looking at your design and at work from Wayne Kirkwood. As far as I could find these are only 2 designs dealing with extreme low signal and R while maintaining excellent CMR... Ok, not only these: Scott Wurcer's design re posted earlier could do, with adopted input stage.
I tried to cook something on my own, but as far I did not get close to yours or Kirkwood design, so I will not show....
Nevertheless you got the point, even dwg was wrong.
Version B Im using for years with 66db gain and 40 ohm cart, there is really no issue neither with overload neither with noise. Few times in thread I tried to point out that high gain = high SNR, and that only limiting factor is overload margin, which I think is overrated.
I'm happy we agree on mic amps, IMO for many DIY-ers this is fastest and very plausible way to get to very good results.
For more extreme LOMC carts, I'm looking at your design and at work from Wayne Kirkwood. As far as I could find these are only 2 designs dealing with extreme low signal and R while maintaining excellent CMR... Ok, not only these: Scott Wurcer's design re posted earlier could do, with adopted input stage.
I tried to cook something on my own, but as far I did not get close to yours or Kirkwood design, so I will not show....
Well, their intended function is clear, isn’t it.
I have more problems with the parallelling of opamps in version C.
They have the same gain, so adding their outputs won’t increase the individual levels, but their uncorrelated noise will be added resulting in a 3dB loss in S/N.
This will only result in a 3dB gain in S/N when a real addition is performed with an extra opamp.
Hans
Hi, seems they are. Sorry for clumsiness , but intention is clear. Unfortunately cant edit original message any more, also no sense to post 3rd sketch. As Hans said, intention is quite obvious.
Correct, there are pit holes in paralleling devices. I'm not advocating circuit C, but it is doable and could work 70sh db SNR @ 66db gain if gains are matched with care. Going much over 66db gain will probably finally get to us to overload margin issues, Besides mic amps go just up to 60db (66db in differential ).
But in this territory (better than 70db SNR) I think better to move to ZTX variations....
Drazen,
I probably didn´t express myself well enough, that's why I made an image with three different options in a simplistic way, but the content should be clear.
When output A has a S/N of Z, then output B will have a S/N of Z-3dB and output C will have a S/N of Z+3dB.
Note that overloading in case of output C is not an issue, because the summing amp through R5 can just as well have a gain of 0.5 or 0.1 or whatever without changing the S/N when input signals are large enough as in your case.
Hans
I probably didn´t express myself well enough, that's why I made an image with three different options in a simplistic way, but the content should be clear.
When output A has a S/N of Z, then output B will have a S/N of Z-3dB and output C will have a S/N of Z+3dB.
Note that overloading in case of output C is not an issue, because the summing amp through R5 can just as well have a gain of 0.5 or 0.1 or whatever without changing the S/N when input signals are large enough as in your case.
Hans
🤔, isn't it what we both say all the time, 2 devices in parallel =3db snr improvement.
Doesn't matter 😊 these are about limits of mic amps, and as we both concluded, even with such limits will be very good solution for most Mc carts.. For some hi out models even superb...
At moment I'm looking at details of that Kirkwood design, some ingenious catches there... At moment have no time to make decent post...
Till later,
Dražen
Doesn't matter 😊 these are about limits of mic amps, and as we both concluded, even with such limits will be very good solution for most Mc carts.. For some hi out models even superb...
At moment I'm looking at details of that Kirkwood design, some ingenious catches there... At moment have no time to make decent post...
Till later,
Dražen
One thing I did not mention in mic amps wrap up is trans-impedance mode, when cart is connected to Rg inputs and +- inputs are grounded.
Both Hierfi and myself had good results while testing that, reducing noise (and THD as it seems) even further.
I don't feel that paralleled devices are wise to use here, so only version A from #693 above is ok. It doesn't have balanced outputs, but input is floating differential and that is what counts most for good CMR. Active diff line driver can always be used after passive RIAA .
However there are limitations coming from the fact that the cart is now actually Rg:
1. dis-balance in Rsource from cart coils will directly make dis-balance in L and R channel gain. I googled a bit on subject of cart coil R measurements, this is not well covered subject at all. Some measurements are taken, guy from Shure stated that up to 5% difference is acceptable (probably referring to MM), someone else said that 10 Ohm is good difference, my expensive Benz has 7% difference (3 Ohm). In any case thing to consider and measure cart R, raw outputs and pream outputs before concluding set up is ok.
2. There is limitation in lowest cart R. Most of these mic amps are not tested for gains above 60db. For SSM2017-2019 that would be Rg= 10 Ohm, so chart R cannot be lower than that, unless we go in unexplored gains over 60db.
THAT1512 has Rf of 2k5 (instead of 5k) which allows for R cart of only 5 Ohm for 60db gain.
Than there is THAT1570 that has all 3 feedback resistors external. Disadvantage is of course that we lose 2 Rf being laser matched on die and need to maliciously match 2 resistors, but advantage is that we can choose Rf 's to our taste , needs for gain and complementing used Rcart.
Cheers,
Dražen
Both Hierfi and myself had good results while testing that, reducing noise (and THD as it seems) even further.
I don't feel that paralleled devices are wise to use here, so only version A from #693 above is ok. It doesn't have balanced outputs, but input is floating differential and that is what counts most for good CMR. Active diff line driver can always be used after passive RIAA .
However there are limitations coming from the fact that the cart is now actually Rg:
1. dis-balance in Rsource from cart coils will directly make dis-balance in L and R channel gain. I googled a bit on subject of cart coil R measurements, this is not well covered subject at all. Some measurements are taken, guy from Shure stated that up to 5% difference is acceptable (probably referring to MM), someone else said that 10 Ohm is good difference, my expensive Benz has 7% difference (3 Ohm). In any case thing to consider and measure cart R, raw outputs and pream outputs before concluding set up is ok.
2. There is limitation in lowest cart R. Most of these mic amps are not tested for gains above 60db. For SSM2017-2019 that would be Rg= 10 Ohm, so chart R cannot be lower than that, unless we go in unexplored gains over 60db.
THAT1512 has Rf of 2k5 (instead of 5k) which allows for R cart of only 5 Ohm for 60db gain.
Than there is THAT1570 that has all 3 feedback resistors external. Disadvantage is of course that we lose 2 Rf being laser matched on die and need to maliciously match 2 resistors, but advantage is that we can choose Rf 's to our taste , needs for gain and complementing used Rcart.
Cheers,
Dražen