A dream came true - R2R.
- Thread starter byrdparis
- Start date
Is there something which documents this project? Clearly the PCB is part of a larger build. Thanks.
Yes, this is all textbook stuff. I wonder if it has occurred to you that your claims apply to transistors too, even opamps, yet somehow they can get some pretty decent numbers out of them (which suggests that something else is afoot...). How much schooling are you asking for here? Do you want me to walk you through it?
One great mind once wrote here:
"tubes benefit from that with regard to CMRR so that value can be decent- well into the 90dB region."
It clearly says: TUBES. There is absolutely no way under the Sun that the tubes can get even close to it. The discussion had to do with tube amplifiers.
For those not familiar with the situation (back to the same textbook stuff again), in order for a circuit to get 90dB of CMRR it must use the resistors with .001% tolerance. When was last time anyone purchased those?
That if you disregard the contribution of active components. Just the basics.
So I would be extremely impressed if a tube differential amplifier had anything higher than, say, 10dB.
"tubes benefit from that with regard to CMRR so that value can be decent- well into the 90dB region."
It clearly says: TUBES. There is absolutely no way under the Sun that the tubes can get even close to it. The discussion had to do with tube amplifiers.
For those not familiar with the situation (back to the same textbook stuff again), in order for a circuit to get 90dB of CMRR it must use the resistors with .001% tolerance. When was last time anyone purchased those?
That if you disregard the contribution of active components. Just the basics.
So I would be extremely impressed if a tube differential amplifier had anything higher than, say, 10dB.
OK: a bit of schooling. There is a simple way to get around the need for high ultra high precision resistors. FWIW, such resistors would be needed for a solid state circuit too so this is not a tube or transistor thing, its a good practice for differential amplifiers.
Here is the problem: Let's take the example of two 5K resistors to ground. A 1 Ohm difference between the two can degrade CMRR performance by as much as 60dB!
So that really isn't the way to set up a differential input although many designers do (although the input balance is not the only way to arrive at a good CMRR value...). To fix that, instead of two resistors to ground off of each input (pin 2 and pin 3 of the XLR) the two resistors instead tie together. A third resistor is then tied to that point and it goes to ground instead. This helps force common mode operation and allows the input resistors to simply be an impedance between the two inputs of the differential amplifier. In this manner matching the two input resistors is far less important. Obviously the higher the value of the common mode resistor, the better the CMRR, within the limits of the input devices, whether tube or solid state.
Hm. It seems as if you ignored most of my last post.
Your closing statement in the post above is false; there's a flaw in your argument. I alluded to it in my last post.
The passive approach then.
I measured 78dB by shorting pins 2 and 3 of the input XLR of our phono section together and driving that and ground. Then drove pins 2 and 3 normally. This kind of test tends to be inaccurate as any difference outside of 0.1% with the input resistors will typically limit the reading to about 60dB.
At the output of the phono with pins 2 and 3 tied together the output was about 0.005mV. When driven normally the output was 40V. This works out to 78dB, which suggests an excellent match of the source resistors. A more sophisticated test and be done (but takes a lot of setup time)which can measure the CMRR more accurately but it will simply show a larger value, in keeping with what I previously stated.
If this circuit were discrete semiconductors it would have a similar CMRR reading for exactly the same reasons, as CMRR has nothing to do with tube or solid state. It has to do with input matching and gain, which is why opamps can obtain high values of CMRR despite poor internal matching.
My statement that balanced circuits will always have higher input noise has been supported by analysis and stands, the "I have never seen this" is not even an argument. The question was how could the single-ended connection even be better, and thus the answer.
Regarding the CMRR, in any serious corporation the designer would be required to provide the worst case analysis of his work, not one unit measurement at one condition. The best GUARANTEED number with .1% resistors is about 50dB, and from that point one should consider the unknown and unspecified tube effect, the temperature coefficient (significant here), and the frequency range. With all those parameters considered the achievable number drops into the 30dB area, if not lower.
It is definitely there, and it definitely helps some, but its effectiveness in a typical home audio product, especially the tube one, is often overestimated.
Furthermore, the CMRR may or may not be of value in a particular system, while increase in noise is bound to affect every one.
I always feel funny when someone is trying to convince me of benefits of balanced topology and interfaces, since my professional background started in electronic instrumentation, but as always the case, the picture is a bit more complicated than some might try to present. And floating sources is one of the poorly understood subjects, where simplistic dogmas don't always apply.
Fortunately, this is audio... "and the living is easy".
Going back to that little project that started this argument, last night I did the listening comparison to a serious contender, and you can hear the result for yourself. The system, with 110dB horns and all tube electronics, is totally quiet.
I am no longer sure how many head preamps I have in the system, but one of my favorites has single-ended input connection. Whether it is responsible for its elegant sound I can't tell, but it surely sounds seductive.
Regarding the CMRR, in any serious corporation the designer would be required to provide the worst case analysis of his work, not one unit measurement at one condition. The best GUARANTEED number with .1% resistors is about 50dB, and from that point one should consider the unknown and unspecified tube effect, the temperature coefficient (significant here), and the frequency range. With all those parameters considered the achievable number drops into the 30dB area, if not lower.
It is definitely there, and it definitely helps some, but its effectiveness in a typical home audio product, especially the tube one, is often overestimated.
Furthermore, the CMRR may or may not be of value in a particular system, while increase in noise is bound to affect every one.
I always feel funny when someone is trying to convince me of benefits of balanced topology and interfaces, since my professional background started in electronic instrumentation, but as always the case, the picture is a bit more complicated than some might try to present. And floating sources is one of the poorly understood subjects, where simplistic dogmas don't always apply.
Fortunately, this is audio... "and the living is easy".
Going back to that little project that started this argument, last night I did the listening comparison to a serious contender, and you can hear the result for yourself. The system, with 110dB horns and all tube electronics, is totally quiet.
I am no longer sure how many head preamps I have in the system, but one of my favorites has single-ended input connection. Whether it is responsible for its elegant sound I can't tell, but it surely sounds seductive.
Last edited:
Citation needed. Extraordinary claims require extraordinary proof.
This is clearly nonsense (it works the other way 'round- click on the link just below). Were it true I would not have been able to measure what I did yesterday.
What you are missing here is that CMRR has little to do with signal symmetry and everything to do with the balance of common mode impedances. If you have technical understanding, then you know this has nothing to do with whether the circuit is tube or solid state.
We measure CMRR to verify CCS operation. As I said, with a a more elaborate method (please note at the link that Analog Devices says something different WRT to your 50dB comment BTW), we would avoid the error of the source resistors and obtain a higher value. Bill Whitlock, the head engineer emeritus of Jensen Transformers, in his writings also seems to disagree with your assertions. You could try this sort of thing yourself- its clear to me that you have not.
Since you worked with instrumentation you should know that all tubes and transistors amplify via differential effect, IOW what is different between their input and the emitter (whether a cathode, actual emitter, source, etc.). Don't believe me? Place a coupling cap between the grid and cathode of a tube and see what happens. The tube amplifies what is different between the grid and ground in a single-ended circuit. By placing a coupling cap between the two inputs, the signal is now common mode so the gain is vastly reduced. The early Philbrick Research (non-production) opamps were pentodes driven diferentially in exactly this manner- one input on the control grid and the other on the cathode.
Could you at least try some of this before trying to gaslight me?
Actually, you would be, and precisely like that. I verified that myself last night.
But I am moving on... this is not a technical forum, and I am sure people are getting bored.
This statement can be interpreted in several ways. It sounds like you are in agreement.
Similar threads
 | Steve Williams Site Founder | Site Owner | Administrator |  | Ron Resnick Site Owner | Administrator |  | Julian (The Fixer) Website Build | Marketing Managersing |