No, I haven't tried it, but I've seen this arrangement in several amps. What's the theory behind it? What improvements are expected?aparatusonitus said:
Yes, but you seem to have a habit of not showing complete schematicslineup said:JFET input AD797
... as close as we can come an ideal amplifier, probably ...
My first simulated 2SK170 JFET input AD797 Clone was some years ago.
I did even post a topic about this
Tremendous results!
. Did you ever build such an amp?
nelsonvandal said:Yes, but you seem to have a habit of not showing complete schematics
Did you ever build such an amp?
Do you want the complete schematic?
I think I can dig it up.
I have explained that right now I have not the practical possibility to build.
And to me this does not matter much.
I leave it to those that need a new amplifier to build.
I hope my circuits can be a good inspiration to ideas.
You do not blame the architect, desginer of houses,
for not do the building himself with his hands, do you?
People are different.
Some try to impose they are more perfect, because they actually build a lot.
They do not tolerate that there are other ways to be.
Theory and praticality are not two enemies.
Anything practical starts as an idea. One can not live without the other.
The eye can not tell the foot, that I am better than you.
Only together we can form full and good body.
Regards
Lineup
Yes, if it's not too much trouble for you.lineup said:
Every contribution, theoretical or practical is what makes a discussion forum interesting. I agree. Some just build and listen to amps, some simulate, some calculate, some just give a hint of their knowledge. As a beginner it can be frustrating not to get everything served in a nice package - theoretical discussion, schematic, pcb, measured performance and listening impressions.
Me too! I haven't put much effort in it yet. I tried it once in a messy unity gain AD829 clone, but it oscillated. I'm going to give it a try again later on.Onvinyl said:
Adding CFP or cascoding the input doesn't alter the simulations to the better. Most often there's no change at all and somtimes it simulates worse, but those tweaks do alter the sound. In other amps, I've found the use of CFP with the combination JFET/bipolar to be an improvement.
jcarr said:
Hi folks,
I found this 3d very interesting.
I have been playing around with topologies a bit and I found the folded cascode with diamond output to play better than the double differential, at least in my set up.
Referring to the post quoted i have a question:
speaking of the shunt transistor T1. Is it possible to run the circuit without it (i.e. by properly and precising setting the CCS)?
I was thinking that this would give then plenty of room to T4 that otherwise would only work with a Vce of about 0.65V
You mean something like the attached schematic? At light loads you may not see much difference in distortion, but the extra circuitry is a benefit with heavier loads.
Of course. However if you do this, distortion will go up due to T4's VCE fluctuations and forward Early voltage. Also if you do this, keeping the current source J7 & RG1 will make the amplifier more difficult to set up, since T4 will no longer be available to absorb deviations in the currents through RE+RF, J3, and J7. IJ3 + IJ7 need to equal IRE + IRF, and with operational temperature fluctuations that may prove difficult. Ergo, if you get rid of T1, also dumping J7 & RG1 will make life easier.
As I stated originally, if you instead add some level shift (voltage) to T1's emitter, you can give T4 more room while still retaining the AD797-ness of Nelson's circuit.
rgds, jonathan carr
Attachments
yeah...i meant for the circuit.
So if understand correctly taking the shunt off is more a disadvantage than an advantage.
I was anyways taking in consideration with my early statement the fact of having all matched devices (monolithic if possible for the front end and for the folded cascode) in order to reduce at minimum fluctation.
Your schemtatic referes to the the output stage though.
at the output i already choose the wildar buffered complementary current mirror with one current source powering the complementary mirrror...instead of having the 2 CCSs.
Speaking of....
1) level shift: you suggested to shif the level of either the base T1 or the emitter.
Which one would be the best accordingly to your experience?
IMHO setting the base voltage at lets say 1V should also help DC stability, is it correct?
2) for the folded cascode, instead of having 2 separates current sources for the LTP of the front end and for the folded cascode, would a nice evolution of a current mirror improve the overall stability and performence of it?
Nevertheless i remember you somewhere mentioning that, for folded cascode where the CCS are very critical, you prefere to use separate current sources cascoded with fixed voltage regulator, am I recalling wrong?
anyways...
3) Do you prefere good ICs for the cascoding or discrete voltage regulator?
Sorry for all these questions, I hope you will find a chance to answer them.
Best.
So if understand correctly taking the shunt off is more a disadvantage than an advantage.
I was anyways taking in consideration with my early statement the fact of having all matched devices (monolithic if possible for the front end and for the folded cascode) in order to reduce at minimum fluctation.
Your schemtatic referes to the the output stage though.
at the output i already choose the wildar buffered complementary current mirror with one current source powering the complementary mirrror...instead of having the 2 CCSs.
Speaking of....
1) level shift: you suggested to shif the level of either the base T1 or the emitter.
Which one would be the best accordingly to your experience?
IMHO setting the base voltage at lets say 1V should also help DC stability, is it correct?
2) for the folded cascode, instead of having 2 separates current sources for the LTP of the front end and for the folded cascode, would a nice evolution of a current mirror improve the overall stability and performence of it?
Nevertheless i remember you somewhere mentioning that, for folded cascode where the CCS are very critical, you prefere to use separate current sources cascoded with fixed voltage regulator, am I recalling wrong?
anyways...
3) Do you prefere good ICs for the cascoding or discrete voltage regulator?
Sorry for all these questions, I hope you will find a chance to answer them.
Best.
Yes, it's quite possible. I went from an AD829 clone to this AD797 clone in steps, and having the floating CM changed it towards better details and brightness, and by adding the shunt transistor the amp sounds smoother. I'm not sure the decompensation cap improved things, as it does with the monolithic AD797.Stefanoo said:
I've lend it to a fellow Head-Fi:er, but when I get it back I'll look in to some of Jonathan's suggestions. The guy I lend it to thinks the amp sounds dark and congested, but I find it very neutral. If I add it to the signal chain it's very hard to hear it's there at all, and like you can see it's "invisible" driving a 33R dummy load on the RMAA tests with a decent soundcard (E-mu 0404). Driving a 10R dummy load, it still looks OK.
I'm looking forward to see your version! I agree, the diamond buffer sounds very good. I've compared it to a single EF, not a double, and I prefer the DB.
I am actually going to use it as a phono stage.
It won't be a line stage.
I was wondering how you managed more voltage across the cascode (did you insert a level shift and verify the sound)?
Did you change the configuration of the current source on the LTP and on the cascode?
I mean you are running 2 current sources. wouldn't it be good to have a good current mirror (i am not talking about the output stage)?
Last, do you guys think it would be possible to achieve with this confiuguration a gain of 65dB?
I would like to have this stage to equalize a 0.35mV MC and up till now i have been using a common source and then this but the output offset with the common source connected it's horrible.
Plus if i could achieve 65dB of gain with just this stage it would be better.
alright....as far as my tests are concerned the folded cascode sounds really balance.
It won't be a line stage.
I was wondering how you managed more voltage across the cascode (did you insert a level shift and verify the sound)?
Did you change the configuration of the current source on the LTP and on the cascode?
I mean you are running 2 current sources. wouldn't it be good to have a good current mirror (i am not talking about the output stage)?
Last, do you guys think it would be possible to achieve with this confiuguration a gain of 65dB?
I would like to have this stage to equalize a 0.35mV MC and up till now i have been using a common source and then this but the output offset with the common source connected it's horrible.
Plus if i could achieve 65dB of gain with just this stage it would be better.
alright....as far as my tests are concerned the folded cascode sounds really balance.
After a long wait I have finally tried a cascoded input stage with the "AD797" curcuit. I've also added some voltage drop to T1 like suggested before. With the added voltage drop there's no longer any problems using this amp at lower supply. I run it from 8 AAA batteries.
I'm not sure if I like the sound of the cascoded input stage better. I have to give it some time. It measures very well and looks good on the scope. The picture shows a 100 kHz square at 2 V (slightly over compensated). The RMAA measurements show to the left E-mu 1212M loopback and right this amp with a 330R load. The 2nd harmonic is slightly raised and the 3rd slightly lowered.
I still have the amp with the standard JFET input stage so I can easily compare them. I also have some other versions. I can say for sure that AD797-izing the standard JFET input amp improves the sound. The sound is cleaner and the spatial information is better. I don't think there are any drawbacks.
I'm not sure if I like the sound of the cascoded input stage better. I have to give it some time. It measures very well and looks good on the scope. The picture shows a 100 kHz square at 2 V (slightly over compensated). The RMAA measurements show to the left E-mu 1212M loopback and right this amp with a 330R load. The 2nd harmonic is slightly raised and the 3rd slightly lowered.
I still have the amp with the standard JFET input stage so I can easily compare them. I also have some other versions. I can say for sure that AD797-izing the standard JFET input amp improves the sound. The sound is cleaner and the spatial information is better. I don't think there are any drawbacks.
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.