Hi, just for info, I added 5n6 over Rg too, setup was still in one piece and took just 2 minutes to add capacitor.
RTA picture remained exactly the same as above in #178. Is bandwidth of soundcard to short to catch differences...
RTA picture remained exactly the same as above in #178. Is bandwidth of soundcard to short to catch differences...
Hi,
I put SSM2017 back in "normal" arrangement: inputs connected at In+ and In-, Rg connections with 33Ohm across and nothing else. All other settings as post #178 (on the scope to right I adjusted only scale to see in and out closer together, nothing to affect result):
On current input setup the noise is just marginally better, for 1.3db (from -88.4 to -87.1 db) . In my opinion this is not worth going into current input mode, especially since in "normal" voltage input mode cartridge loading can be freely adjusted and there is no need for input DC bias adjustment.
What I thing can be concluded according to my and Hierfi observations , and fact that Calvin and myself are using those for decades without issues, is that SSM2017 (2019) are very competent chips and performing better than specifications.
These are neither opamps neither INA (even block diagram from outside looks like INA), but dedicated differential in, current feedback preamps, spices on their own.
Since time is getting closer when I will receive my TT parts, next thing I will do is clean my desk and take TT apart . Then I can quickly make RTA of fully differential first MC stage that I was using and being very happy with...
I put SSM2017 back in "normal" arrangement: inputs connected at In+ and In-, Rg connections with 33Ohm across and nothing else. All other settings as post #178 (on the scope to right I adjusted only scale to see in and out closer together, nothing to affect result):
On current input setup the noise is just marginally better, for 1.3db (from -88.4 to -87.1 db) . In my opinion this is not worth going into current input mode, especially since in "normal" voltage input mode cartridge loading can be freely adjusted and there is no need for input DC bias adjustment.
What I thing can be concluded according to my and Hierfi observations , and fact that Calvin and myself are using those for decades without issues, is that SSM2017 (2019) are very competent chips and performing better than specifications.
These are neither opamps neither INA (even block diagram from outside looks like INA), but dedicated differential in, current feedback preamps, spices on their own.
Since time is getting closer when I will receive my TT parts, next thing I will do is clean my desk and take TT apart . Then I can quickly make RTA of fully differential first MC stage that I was using and being very happy with...
Hi Bonsai,
My turntable is waiting for motor drive parts, it will take another month or so before I can test with TT.
With signal generator, mimicking cart with 33 Ohm, 0,34mV @ 3,54 cm/s (that's how they specified, not 5cm/s) and 1kHz, SNR is -75,9 db
My turntable is waiting for motor drive parts, it will take another month or so before I can test with TT.
With signal generator, mimicking cart with 33 Ohm, 0,34mV @ 3,54 cm/s (that's how they specified, not 5cm/s) and 1kHz, SNR is -75,9 db
As an aside Bonsai, in your measurements of noise do you apply post RIAA equalization to evaluate the noise spectrum and a single figure of noise measurement?
Not at the moment, I'm just playing with 1st stage now.
To get complete preamp after this comes passive riaa that will reduce mids by 20 and HF by 40db, than something else to amplify whole fr again to final output level
To get complete preamp after this comes passive riaa that will reduce mids by 20 and HF by 40db, than something else to amplify whole fr again to final output level
Hi,
For the moment I would leave ssm2017(9) at peace these are obviously very competent preamps, and instead come back to discrete solutions.
Since quite some time I'm contemplating about universal FDA, good circuit that has differential in and out and can be applied (as raw diagram, not as all details) to all fully differential audio, from MC head amps, via preamps, mixers to power amp voltage stages.
Back to the subject of this thread, there are many MC head preamp configurations intended to plug in MM preamp, somewhere earlier I argued that it is better to have much more gain than that for the first stage, which only few configurations can do, and even fewer that are FDA.
This type of circuit needs to have feedback, and if we picture opamp, than it is 4 inputs and 2 outputs. With opamp it is as easy as this:
But it comes with opamp limitations:
Then there are ready made FDA-s, again with limits:
Then INA-s and specialized SSM preamps, well, they integrate differential outputs to single ended so 2 are necessary (for differential output), otherwise all the same as opamp solution.
I use this as MC preamp as in post #32, as well as opamp solution for second stage, it is very good and appropriate, with all limitations of above.
Contemplation of ideal FDA led me to this:
Basic CFA , 2 of them connected in LTP to make FDA. This is kind of simplified input circuit of SSM2017, but if done discretely it can be done to taste and need, all components and PS can be adopted for whatever one wants.
This is what I see as optimal, but did not built it yet,,, I have to now.
Second idea I already used in quite large but simple poweramp long time ago, it is described here https://www.diyaudio.com/community/...ttle-fb-160watt-amplifier.412261/post-7672583
Same idea is in Broskie's diagram I copied in post#1 of this thread.
LTP with second stage and inverting FB configuration. This configuration is proven but it still suffers from one thing: Zin cant be increased without sacrificing noise. I just made second setup of Broskie's diagram and will post it immediately.
Critics and opinions appreciated,
Cheers,
Drazen
For the moment I would leave ssm2017(9) at peace these are obviously very competent preamps, and instead come back to discrete solutions.
Since quite some time I'm contemplating about universal FDA, good circuit that has differential in and out and can be applied (as raw diagram, not as all details) to all fully differential audio, from MC head amps, via preamps, mixers to power amp voltage stages.
Back to the subject of this thread, there are many MC head preamp configurations intended to plug in MM preamp, somewhere earlier I argued that it is better to have much more gain than that for the first stage, which only few configurations can do, and even fewer that are FDA.
This type of circuit needs to have feedback, and if we picture opamp, than it is 4 inputs and 2 outputs. With opamp it is as easy as this:
But it comes with opamp limitations:
- max V is limited by opamp, good just for smaller power amps
- min noise is limited to extend of AD797
- me like to avoid chips (not really, but lets assume) and do things own way 🙂
Then there are ready made FDA-s, again with limits:
- all the same as above for opamps
- additionally, these are by default inverting feedback devices, resulting in low Zin. This is ok for MC preamp, but bad for MM or second phono stage after passive RIAA or buffer after the potentiometer for example.
Then INA-s and specialized SSM preamps, well, they integrate differential outputs to single ended so 2 are necessary (for differential output), otherwise all the same as opamp solution.
I use this as MC preamp as in post #32, as well as opamp solution for second stage, it is very good and appropriate, with all limitations of above.
Contemplation of ideal FDA led me to this:
Basic CFA , 2 of them connected in LTP to make FDA. This is kind of simplified input circuit of SSM2017, but if done discretely it can be done to taste and need, all components and PS can be adopted for whatever one wants.
This is what I see as optimal, but did not built it yet,,, I have to now.
Second idea I already used in quite large but simple poweramp long time ago, it is described here https://www.diyaudio.com/community/...ttle-fb-160watt-amplifier.412261/post-7672583
Same idea is in Broskie's diagram I copied in post#1 of this thread.
LTP with second stage and inverting FB configuration. This configuration is proven but it still suffers from one thing: Zin cant be increased without sacrificing noise. I just made second setup of Broskie's diagram and will post it immediately.
Critics and opinions appreciated,
Cheers,
Drazen
Just FYI, I made another run on Broskie's idea test rig, this time little bit more elaborated:
In above, red pen are changes and additions exactly as tested.
This diagram is same principle as idea B in previous post, just upside down with PNP instead opf NPN transistors, and simplified without current sources.
This is the result with T1 = SSM2220:
Guess not bad for few components soldered on the fly. PS is only 2x 9V battery, no regulators or capacitors.
This setup is beating SSM2017 on noise, but not on THD. Both are no wonder. Notice is that SSM2017 I run with higher gain of 50db which should yield opposite results (lower noise and higher THD).
I repeated the same with unmatched ZTX951 and noise was just slightly better, distortion slightly worse (no wander since they are not matched). Probably ZTX will show its noise competence with real cart instead of high value resistors reducing signal, but that will wait till I set TT.
I conducted one more run just to get more confidence in setup , this time with some old BC160 transistors as T1, the result was much worse in all aspects.
These last 2 pictures I'm not posting just to reduce size of data for admins.
In above, red pen are changes and additions exactly as tested.
This diagram is same principle as idea B in previous post, just upside down with PNP instead opf NPN transistors, and simplified without current sources.
- Resistors to bases of T1's , 1R and 100R form inverting feedback loop for gain of 40db.
- 33R is mimicking cartridge
- I brought the signal (1kHz - 1VRMS) over two 7k5 resistors, calculation gives 0.13 mV input voltage as seen at cart input, I measured 0.19 mV, but that is within tolerances.
- For T2 I always used SSM2220
- For T1 I switched between SSM2220 and ZTX915 (unmatched, just from the shop)
- This measurement is just indication; it is very difficult to mimic MC low Z and low output, in addition it is just linear amp, no RIAA or second stage.
This is the result with T1 = SSM2220:
Guess not bad for few components soldered on the fly. PS is only 2x 9V battery, no regulators or capacitors.
This setup is beating SSM2017 on noise, but not on THD. Both are no wonder. Notice is that SSM2017 I run with higher gain of 50db which should yield opposite results (lower noise and higher THD).
I repeated the same with unmatched ZTX951 and noise was just slightly better, distortion slightly worse (no wander since they are not matched). Probably ZTX will show its noise competence with real cart instead of high value resistors reducing signal, but that will wait till I set TT.
I conducted one more run just to get more confidence in setup , this time with some old BC160 transistors as T1, the result was much worse in all aspects.
These last 2 pictures I'm not posting just to reduce size of data for admins.
Sorry, quoting myself, Correction. SSM2017 is so far little better on noise too, just for this ad hook setup.
Another option is to feed the SSM2017/19 into a drv135 single ended to differential out. Keep in mind that the SSM devices have a "sense" pin that can remote sense the ground at the drv135 input ground pin, hence doesn't use the power supply ground returns. The connection would be to an output XLR would connect the sense pin to the negative in the XLR, being an isolated "clean" ground to a remote DRV135.
https://www.ti.com/lit/ds/symlink/drv135.pdf?ts=1722111264519&ref_url=https%3A%2F%2Fwww.google.com%2F
Last edited:
Hi, I agree , that would work for sure, typical pro audio engineer thinking. But I don't see how is it better to put 2 devices in series comparing to put 2 SSM2017-9 in parallel? Mc chart will have no issues driving double input current..
One reservation; I have to admit that I never used DRV chips, but I did SSM2142 and I believe they are very similar if not the same, like 2017 and 2019..
Then there is the THAT 1646 balanced line driver and the 12xx series of balanced line receivers.
There is an interesting thread below on this topic, different driver topologies, and using multiple1646's with build out resistors to drive very long lines.
https://www.proaudiodesignforum.com/forum/php/viewtopic.php?f=6&t=635
There is an interesting thread below on this topic, different driver topologies, and using multiple1646's with build out resistors to drive very long lines.
https://www.proaudiodesignforum.com/forum/php/viewtopic.php?f=6&t=635
Hello,
In the meantime I made one more rig and run it, this time proposal A from post #188 above, just I turned it around and used PNP at input and NPN at output; hence letter A upside down. In principle this is similar to SSM2017 circuit but much simplified:
For thze test purposes I skipped current sources and installed 4 4k42 resistors that gave approx 2 mA bias per transistor.
Input was 1V RMS from soundcard, run through 47k resistors, that gave approx 0.35 mV across 33R that was playing role of a cart.
T1 was ZTX951 (later switched for SSM2220) and T2 was always ZTX851. (ZTX-es not matched!)
Here an idea popped up; can we, did anyone made it like this, put HF filter pole of RIAA straight here on tails of first LTP?
If this is plausible, all overload questions would vanish as HF would be cut straight at the very beginning of amplification.
I think I have seen it done like that somewhere, but cant find it anymore... 🤔
Here is test result, (without red circle filter):
I also tried swapping T1 with SSM2220; the noise was slightly worse, but THD better (2nd order almost disappeared in noise). Similar like in earlier posts when doing the same swap, here:
Than I run it in trans-impedance mode; shortened present inputs to ground and brought the signal to emitters of T1's via 1k resistors. The gain is now approx x10 higher as I did not change 4R7 :
The noise is now really nice at -79.7db SNR, but distortion went up a lot!
In the meantime I made one more rig and run it, this time proposal A from post #188 above, just I turned it around and used PNP at input and NPN at output; hence letter A upside down. In principle this is similar to SSM2017 circuit but much simplified:
For thze test purposes I skipped current sources and installed 4 4k42 resistors that gave approx 2 mA bias per transistor.
Input was 1V RMS from soundcard, run through 47k resistors, that gave approx 0.35 mV across 33R that was playing role of a cart.
T1 was ZTX951 (later switched for SSM2220) and T2 was always ZTX851. (ZTX-es not matched!)
Here an idea popped up; can we, did anyone made it like this, put HF filter pole of RIAA straight here on tails of first LTP?
If this is plausible, all overload questions would vanish as HF would be cut straight at the very beginning of amplification.
I think I have seen it done like that somewhere, but cant find it anymore... 🤔
Here is test result, (without red circle filter):
I also tried swapping T1 with SSM2220; the noise was slightly worse, but THD better (2nd order almost disappeared in noise). Similar like in earlier posts when doing the same swap, here:
Than I run it in trans-impedance mode; shortened present inputs to ground and brought the signal to emitters of T1's via 1k resistors. The gain is now approx x10 higher as I did not change 4R7 :
The noise is now really nice at -79.7db SNR, but distortion went up a lot!
Ill try to summarize what I tried in last days:
Valid for all tests:
Single SSM2017 (diff In, single out) in "normal" voltage mode:
SNR: 77db THD: 0.0012% Gain x 303 v/V
Single SSM2017 in current (bus summing) mode:
SNR: 77.3db THD: 0.0012% Gain x 303V/V
Version B (post #188) or inverting FB amp alla Tubecad:
SNR: 73.2db THD: =0.0057% Gain about X100V/V
Version A (post above) in voltage mode:
SNR: 62.7db THD: =0.028% Gain about X100V/V
Version A in current (trans-impedance) mode:
SNR: 79.7db THD: =0.64% Gain about X100V/V
I will not pretend that I can interpret above properly, but what it seems to me is:
Please do not take above as my statement of definitive science proof, just as observations of empiric tests conducted with all limitations as shown.
Please also comment and feel free to spit if you see that I missed point anywhere 🙂 .
Valid for all tests:
- PS is 9V battery, single or double, no additional PS filtering or regulation
- All run through RME Fireface. Signal generator from RME at approx 1 V reduced by resistors to match cart output
- All in metal box with decent shielded differential connections
- All hardwired as found appropriate. (no particular layout care taken.
- PNP = ZTX951 (compared to SSM2220 sometimes), NPN=ZTX851
- In all tests this is linear one gain stage, no RIAA or complete phono preamp yet
- Test bandwidth 20 - 20,000Hz
Single SSM2017 (diff In, single out) in "normal" voltage mode:
SNR: 77db THD: 0.0012% Gain x 303 v/V
Single SSM2017 in current (bus summing) mode:
SNR: 77.3db THD: 0.0012% Gain x 303V/V
Version B (post #188) or inverting FB amp alla Tubecad:
SNR: 73.2db THD: =0.0057% Gain about X100V/V
Version A (post above) in voltage mode:
SNR: 62.7db THD: =0.028% Gain about X100V/V
Version A in current (trans-impedance) mode:
SNR: 79.7db THD: =0.64% Gain about X100V/V
I will not pretend that I can interpret above properly, but what it seems to me is:
- SSM2017 is hard cookie to beat
- Trans-impedance mode brings noise benefits (circuit A) but at cost of THD. With SSM which is high gain-bandwidth device, there is no much difference.
- ZTX951 is just slightly more quiet than SSM2220 (previous post) or SSM2017 as whole block, and that only in trans-impedance mode. All are most likely quiet enough for purpose, LP-stylus noise will bury all of those noises IMO.
Please do not take above as my statement of definitive science proof, just as observations of empiric tests conducted with all limitations as shown.
Please also comment and feel free to spit if you see that I missed point anywhere 🙂 .
Hello all,
I'm cleaning my desk from preamp tests as parts for Turntable rebuilt are gathering. Continuation will be here: https://www.diyaudio.com/community/...eded-with-motor-and-drive.412984/post-7687982
When Im done and happy with TT speed and drive, I will come back to this but next time trying and testing MC phono preamps with real source from TT and test record. That will be late September as in few weeks we are gone fishing for holidays.
This is just a picture of my "Spiders from the Mars" created in course of this chat.
I'm cleaning my desk from preamp tests as parts for Turntable rebuilt are gathering. Continuation will be here: https://www.diyaudio.com/community/...eded-with-motor-and-drive.412984/post-7687982
When Im done and happy with TT speed and drive, I will come back to this but next time trying and testing MC phono preamps with real source from TT and test record. That will be late September as in few weeks we are gone fishing for holidays.
This is just a picture of my "Spiders from the Mars" created in course of this chat.
Last edited:
With minimal 3nV/rtHz, noise will be on the high side for the 0.15mV Cart that is mentioned in the diagram.
The other thing is that Cart termination isn’t well defined with 2*50R typical but without further details on frequency dependency which could affect frequence response.
Hans
The other thing is that Cart termination isn’t well defined with 2*50R typical but without further details on frequency dependency which could affect frequence response.
Hans
U1 and U2 lack negative feedback. They lack bias current path for their inverting inputs. They are also significantly more noisy than the AD797's that follow them - why not use AD797's in the input stage? Why not use the standard instrumentation amp topology come to that?
Thanks for your attention, comment and question. A standard instrument amplifier is a very good thing in other cases. Here we use the Common Bases as an I/V converter - because the MC cartridge is like a current source.
And for some, the lack of feedback is a must!
The current noise of the AD844 is negligible. Especially in a balanced topology?
And the slew rate is uniquely high (fast)!......
My previous project was with AD797 at the first stage!
I used its inverting inputs for the signal from the MC cartridge. It wasn't bad and more like an Instrumental Amp, but the inevitable feedback loop "ate" from the beauty and transient processes of the Input Signal!
And for some, the lack of feedback is a must!
The current noise of the AD844 is negligible. Especially in a balanced topology?
And the slew rate is uniquely high (fast)!......
My previous project was with AD797 at the first stage!
I used its inverting inputs for the signal from the MC cartridge. It wasn't bad and more like an Instrumental Amp, but the inevitable feedback loop "ate" from the beauty and transient processes of the Input Signal!