Hi Miniwatt,
Some correction to what I said above.
If I understand you right there is a remote controlled volume pot of 100k at the input of your power amp since the FVP5 makes use of a 50k volume pot yes/no? And in that case you do not need to put in a complete new voltage divider into the power amp with the values I gave you before because a volume pot is nothing else then a variable voltage divider. Then indeed you can simply add some resistance between the input of the pot and the cinch jack assumed that there is no other resistor present between the pot and the cinch of course. I would buy some pairs of 1% tolerance small metal film resistors with let's say 20k, 10k , 5k, 2k and 1k. Would start with 10k and see if you can get a smoother jump with this and if not increase or decrease resistance accordingly. That's at least what I would try.
Regards,
Günter
Some correction to what I said above.
If I understand you right there is a remote controlled volume pot of 100k at the input of your power amp since the FVP5 makes use of a 50k volume pot yes/no? And in that case you do not need to put in a complete new voltage divider into the power amp with the values I gave you before because a volume pot is nothing else then a variable voltage divider. Then indeed you can simply add some resistance between the input of the pot and the cinch jack assumed that there is no other resistor present between the pot and the cinch of course. I would buy some pairs of 1% tolerance small metal film resistors with let's say 20k, 10k , 5k, 2k and 1k. Would start with 10k and see if you can get a smoother jump with this and if not increase or decrease resistance accordingly. That's at least what I would try.
Regards,
Günter
Last edited:
Not known, but later we discussed using Darlington pairs, low noise matched bipolars. The idea came from Rowan McCombe before he passed away, better known as the guru in Allen's book. You guys might like to try it. Just be sure to match hfe and the 2SC2547 was a favourite of Allen's.
Value 2 x 22K may be tried or left out altogether. Vref for the CS is a TL431 and the "Bias" resistor for it would be typically 15K. The AC Trim may be replaced with two fixed resistors as AC balance should be fine if the Darlington pairs are well matched. The Bias voltage for the 6922a/b and CS I would use +20V, but +15V would likely be OK too.
Value 2 x 22K may be tried or left out altogether. Vref for the CS is a TL431 and the "Bias" resistor for it would be typically 15K. The AC Trim may be replaced with two fixed resistors as AC balance should be fine if the Darlington pairs are well matched. The Bias voltage for the 6922a/b and CS I would use +20V, but +15V would likely be OK too.
Hi Joe,
Very interesting! What value would be needed for the resistors in the CS and the trimmers ( both AC Trim and trimmer at the right hand side of the IRF820 ) and even more important what is the gain of this circuit? Is it comparable with MAT02 or slightly lower?
Thanks, Günter
Very interesting! What value would be needed for the resistors in the CS and the trimmers ( both AC Trim and trimmer at the right hand side of the IRF820 ) and even more important what is the gain of this circuit? Is it comparable with MAT02 or slightly lower?
Thanks, Günter
Last edited:
Hi Jiri,
Regarding your questions w.r.t. the phono input stage I would say it depends on what kind of MC cartridge you wanna use with it. If it is a medium to high output MC cart then the Jfets might be good enough if you want to use it with a low or very low output MC cartridge I would say the solution with the MAT02 is more appropriate since it will provide 12 dB more gain and probably lower noise. From practical considerations the two J-Fets ( 2SK170 or 369 ) must be tightly matched of course and glued together for good temperature balance. Looking at that a MAT02 or SSM2212 is more practical to use since it is already tightly matched and the devices are in one package and as such easier to handle. In terms of sonics Allen clearly preferred the solution with the bipolar transistors, question of taste of course and the new circuit Joe shows here above underlines that I think.
Cheers,
Günter
Regarding your questions w.r.t. the phono input stage I would say it depends on what kind of MC cartridge you wanna use with it. If it is a medium to high output MC cart then the Jfets might be good enough if you want to use it with a low or very low output MC cartridge I would say the solution with the MAT02 is more appropriate since it will provide 12 dB more gain and probably lower noise. From practical considerations the two J-Fets ( 2SK170 or 369 ) must be tightly matched of course and glued together for good temperature balance. Looking at that a MAT02 or SSM2212 is more practical to use since it is already tightly matched and the devices are in one package and as such easier to handle. In terms of sonics Allen clearly preferred the solution with the bipolar transistors, question of taste of course and the new circuit Joe shows here above underlines that I think.
Cheers,
Günter
Hi Joe,
Forgot one question to ask. Specified current through each side of Allen's latest phono cascode with MAT02/SSM2210 was 9 mA, so in total 18 mA to my knowledge. Circuit can be found here under #470:
https://www.diyaudio.com/community/threads/mpp.154210/page-24
Is this here the same with the Darlington pairs of 2SC2547 or do they require a higher current considering the fact that the CS circuit makes use of an IRF820?
Thanks again,
Günter
Forgot one question to ask. Specified current through each side of Allen's latest phono cascode with MAT02/SSM2210 was 9 mA, so in total 18 mA to my knowledge. Circuit can be found here under #470:
https://www.diyaudio.com/community/threads/mpp.154210/page-24
Is this here the same with the Darlington pairs of 2SC2547 or do they require a higher current considering the fact that the CS circuit makes use of an IRF820?
Thanks again,
Günter
Last edited:
Hi Günter,
it seems to me all is about first) high slew rate, second) gain of the input cascode.
As Joe posted above darlington version, it sure have high chance to fulfill both... nevertheless darlington as 2 BJT stages instead one, how about phase shift, and high input impedance impact to noise (even phono cartridge is low impedance source)?
To your reply, I have now and inted to keep Hi output MC (now using ClearAudio MC Essence and I am planning some upgrade, but we are still talking about cca 0.4mV MC output); now I use MAT02 so will keep it
I was just interrested if there is any "typical" sonic influence using j-fet instead BJT, if I consider Allen's statement that lower half of cascode has no sonical imprint because of low voltage gain..
thanks, Jiri
it seems to me all is about first) high slew rate, second) gain of the input cascode.
As Joe posted above darlington version, it sure have high chance to fulfill both... nevertheless darlington as 2 BJT stages instead one, how about phase shift, and high input impedance impact to noise (even phono cartridge is low impedance source)?
To your reply, I have now and inted to keep Hi output MC (now using ClearAudio MC Essence and I am planning some upgrade, but we are still talking about cca 0.4mV MC output); now I use MAT02 so will keep it
I was just interrested if there is any "typical" sonic influence using j-fet instead BJT, if I consider Allen's statement that lower half of cascode has no sonical imprint because of low voltage gain..thanks, Jiri
Hi Jiri,
I'm wondering where you found this "if I consider Allen's statement that lower half of cascode has no sonical imprint because of low voltage gain.."? Remember that the use of a MAT02 in place of a FET gives already 12 dB more gain. Low voltage gain???
In his article Secrets of the phono stage Allen indeed said on page 3 that a FET used in the bottom half of a cascode is sonically completely invisible but that was mainly addressed to silicon haters which sneer when seeing any silicon in tube circuits:
http://www.vacuumstate.com/fileupload/SP_15_Article.pdf
However on the other hand in the kit instructions of the phono stage he said "If even more gain is needed, the jfets can be replaced with a dual monolithic bipolar transistor such as the LM394 or the MAT02—with no other circuit or value changes—which will provide at least another 12dB. This will change the sound however, and you may (or may not) prefer it. With the LM394 I hear even more definition".
So a bit inconsistent in my opinion if you ask me to say no sonical imprint when using transistors no matter if FET's or BJT's in the bottom half of a cascode.
Regards,
Günter
I'm wondering where you found this "if I consider Allen's statement that lower half of cascode has no sonical imprint because of low voltage gain.."? Remember that the use of a MAT02 in place of a FET gives already 12 dB more gain. Low voltage gain???
In his article Secrets of the phono stage Allen indeed said on page 3 that a FET used in the bottom half of a cascode is sonically completely invisible but that was mainly addressed to silicon haters which sneer when seeing any silicon in tube circuits:
http://www.vacuumstate.com/fileupload/SP_15_Article.pdf
However on the other hand in the kit instructions of the phono stage he said "If even more gain is needed, the jfets can be replaced with a dual monolithic bipolar transistor such as the LM394 or the MAT02—with no other circuit or value changes—which will provide at least another 12dB. This will change the sound however, and you may (or may not) prefer it. With the LM394 I hear even more definition".
So a bit inconsistent in my opinion if you ask me to say no sonical imprint when using transistors no matter if FET's or BJT's in the bottom half of a cascode.
Regards,
Günter
hi Günter, you are right, I miss that notices...
I thought that cascode with BJT has higher gain as a whole, due to BJT's beta and thus higher current changes feeding upper triode's cathode input - and then it is the upper triode which gives higher voltage gain comparing using J-fet.
By the way, I also tested LM394 before MAT02 - sound differences between the two seemed to me relatively negglible, LM more with "body", MAT more delicate at highs.
Jiri
I thought that cascode with BJT has higher gain as a whole, due to BJT's beta and thus higher current changes feeding upper triode's cathode input - and then it is the upper triode which gives higher voltage gain comparing using J-fet.
By the way, I also tested LM394 before MAT02 - sound differences between the two seemed to me relatively negglible, LM more with "body", MAT more delicate at highs.
Jiri
The CS is only a suggestion, that part could be done in a number of different ways. What is shown as IRF820 and reference by TL431, has the advantage, if set up right, can handle fairly high negative voltage.
The current can be anything from 4-5mA and up to 9-10mA per phase (up to 20mA in total if you want to push it), the Anode resistor value you would need to figure out depending on the current, but 25-33K seems to be about right. The gain is very high and probably some local degeneration resistors via emitters to CS. The 22K may best be left out, but you can adjust and report the sound here.
If the stage is used as a front-end to a tube output stage and output transformers, then you must leave the 22K out as it tends to impede very large voltage swings. The B+ will likely need to be 300V, but with around 5-6mA current (10-12mA CS) and 25K, you might get it down to 250V. If used in a power amp, the voltage needs to be higher and I would use 33K. You just have to use common sense, depending on what you want to do.
That should cover it so far?
Cheers, Joe
The current can be anything from 4-5mA and up to 9-10mA per phase (up to 20mA in total if you want to push it), the Anode resistor value you would need to figure out depending on the current, but 25-33K seems to be about right. The gain is very high and probably some local degeneration resistors via emitters to CS. The 22K may best be left out, but you can adjust and report the sound here.
If the stage is used as a front-end to a tube output stage and output transformers, then you must leave the 22K out as it tends to impede very large voltage swings. The B+ will likely need to be 300V, but with around 5-6mA current (10-12mA CS) and 25K, you might get it down to 250V. If used in a power amp, the voltage needs to be higher and I would use 33K. You just have to use common sense, depending on what you want to do.
That should cover it so far?
Cheers, Joe
Hi Joe,
Thank you very much for your answers.
If I maybe do this mod I will apply it in Allen's latest version of the RTP3 phono stage ( link to schematic given above under #806 ) as this is the easiest way by replacing T1 and T2 ( SSM2210 ) with these Darlington's. This circuit makes use of separate current sources for inverted and non-inverted phases around the 2SK170 Fet's and they are set to 9mA each, so ok. But of course they can also be made adjustable by replacing
R20/21 ( 1k3 ) with trimmers so AC Trim in your schematic can be omitted. Anode resistors and B+ voltage are also correct in Allen's circuit, so this modification can relatively easy be done I guess.
Only thing still visible there is R1 instead of your AC Trim and with this resistor the gain of the circuit can be changed. According to the kit instructions value can be set to 0 Ohm for MC or 634 Ohm for MM so this value must be checked of course with the Darlington's in order to get the right gain I would assume.
Cheers, Günter
Thank you very much for your answers.
If I maybe do this mod I will apply it in Allen's latest version of the RTP3 phono stage ( link to schematic given above under #806 ) as this is the easiest way by replacing T1 and T2 ( SSM2210 ) with these Darlington's. This circuit makes use of separate current sources for inverted and non-inverted phases around the 2SK170 Fet's and they are set to 9mA each, so ok. But of course they can also be made adjustable by replacing
R20/21 ( 1k3 ) with trimmers so AC Trim in your schematic can be omitted. Anode resistors and B+ voltage are also correct in Allen's circuit, so this modification can relatively easy be done I guess.
Only thing still visible there is R1 instead of your AC Trim and with this resistor the gain of the circuit can be changed. According to the kit instructions value can be set to 0 Ohm for MC or 634 Ohm for MM so this value must be checked of course with the Darlington's in order to get the right gain I would assume.
Cheers, Günter
Hi guys,
after rethinking your comments related to SLCF output stage which I involved in my RTP3 build (triode as CS and FET at position of active bootstrap, which I finally disconnected and leave this FET fully open) I made several simulations of circuits with FET at CS position and triode udes as bootstrap unit.
I could not duplicate my issue with oscillations of FET bootstrap, and found that ECC88 anf FET shows very similar results at simulations (using SIMetrix sw) - may be FET little better claimed bandwidth for both positions CS/bootstrap and slihgtly higher calculated distortion then tube at CS position. Even ringing on rising/falling edge of square signal was seen very comparable between FET and tube both resistive and small-C loads.
..for me interresting results.
Therefore I decided to leave tube as CS position and FET above triode-CF, but instead bootstraping it to get constant voltage across tube, I just lowered tube's B+ voltage by constant FET voltage drop. Simulations showed lowest calculated distortion (simulated whole preamp cascode+SCLF) @ approx. 55V across CF-ECC88 (before I have nearly 130V here). Values was 0.04% before and 0.26% now.
So I rebuilt SCLF as described here, which was relative very easy mod and listen results..... It seem to me the sound is discernable smoother and more relaxed than before but still very involving and clean - good value for such small mod
There is no bootsraping function at SCLF - it means no any ringing source with clean rising/falling edges,
Jiri
after rethinking your comments related to SLCF output stage which I involved in my RTP3 build (triode as CS and FET at position of active bootstrap, which I finally disconnected and leave this FET fully open) I made several simulations of circuits with FET at CS position and triode udes as bootstrap unit.
I could not duplicate my issue with oscillations of FET bootstrap, and found that ECC88 anf FET shows very similar results at simulations (using SIMetrix sw) - may be FET little better claimed bandwidth for both positions CS/bootstrap and slihgtly higher calculated distortion then tube at CS position. Even ringing on rising/falling edge of square signal was seen very comparable between FET and tube both resistive and small-C loads.
..for me interresting results.
Therefore I decided to leave tube as CS position and FET above triode-CF, but instead bootstraping it to get constant voltage across tube, I just lowered tube's B+ voltage by constant FET voltage drop. Simulations showed lowest calculated distortion (simulated whole preamp cascode+SCLF) @ approx. 55V across CF-ECC88 (before I have nearly 130V here). Values was 0.04% before and 0.26% now.
So I rebuilt SCLF as described here, which was relative very easy mod and listen results..... It seem to me the sound is discernable smoother and more relaxed than before but still very involving and clean - good value for such small mod
There is no bootsraping function at SCLF - it means no any ringing source with clean rising/falling edges,
Jiri
All you are doing is reducing the voltage seen by the ECC88 follower. Other than that, it seems to have zero function. The whole purpose of SLCF (which I BTW was one who coined it and Allen liked it) is that it keeps the follower in near constant voltage and constant current. You have lost the constant voltage function and it is by definition no longer SLCF.
Yes, FET here only keeps constant voltage drop for lower Ua (not regulated but tied to B+ rail because input cascode stage and its direct coupling to CF), so there is no constant Eak of CF triode.All you are doing is reducing the voltage seen by the ECC88 follower. Other than that, it seems to have zero function. The whole purpose of SLCF (which I BTW was one who coined it and Allen liked it) is that it keeps the follower in near constant voltage and constant current. You have lost the constant voltage function and it is by definition no longer SLCF.
I mean it as a simple and relative good compromise to achieve more touching sound than ordinary CF with hopefully less distortion. Bootstrapping CF function is still avoided as I see here possible oscilation/ringing source.
I will call this mod let say SFCF - super friendly cathode follower
Apologies if I have already said this, but when I was discussing my SLCF-based active crossover with Allen a few years ago, he told me that if you wanted to replace one of the 6922s with a MOSFET it should be the lower one (the constant current sink), as - in his opinion - the gate capacitance of the FET here adds less of a signature to the sound than in the bootstrap position.Yes, FET here only keeps constant voltage drop for lower Ua (not regulated but tied to B+ rail because input cascode stage and its direct coupling to CF), so there is no constant Eak of CF triode.
I mean it as a simple and relative good compromise to achieve more touching sound than ordinary CF with hopefully less distortion. Bootstrapping CF function is still avoided as I see here possible oscilation/ringing source.
I will call this mod let say SFCF - super friendly cathode follower
When I was starting to build my RTP3C I looked very briefly at the RTP5, with its lower valve count, but came to the decision very quickly that it was a no-brainer to go straight to the all-valve RTP3 version.
Alex
Yes Alex I have seen yours and Gunter's statements above - this is a reason why I again opened preamp's cover and play with it - and enjoyed it
But as I wrote, I am limited with this RTP3 build to max 4 ECC88 tubes per channel, so I must use FET here (or use nothing).
Unfortunatelly my experience is that an oscillations/ringing are present when FET is used for bootstrap, and stable CF stage when no bootstrap is avoided. So I have no any listenning experience using SLCF at my RTP3.
All my other "experiences" are based only by simulating circuits with concern namely to calculated bandwith, response of square signal, and distortion. On the other hand, simulations always lead me to better real results. Sure I can not say (at this case) that ECC88 (SF)CF will best performing at Eak=55V or 50V or 60V, and real distortion will be 0.026% or 0.03% or 0.05% etc., but sure will better than common CF.
And my simulated "experience" using tube at bootstrap position and FET as CS did not show some remarkable calculated improvement. As I wrote above, FET as CS gives slightly better calculated bandwidth, but only of few percent, still ringing on edges - I have to simulate gate/grid resistor of 50-150k to dump ringing (when either tube or FET bootstrap) to achieve reasonable simuated results, and then again both tube or FET do not achieve any significant calculated improvement comparing to CF.
(Simulation of 120k dumpimg resistor at this case correlates well with my real experience before I finally disconnected bootsrapping FET when I played first time with my RTP3 build).
Therefore based only on simulations, I have decided to leave tube as CS and try to incorporate FET "above CF tube" some other way - if it is already placed at my PCBs.
So my point is, this mod was very easy and free of charge, while sonical results seems to improved.
If real distortion decrased of half or third value from usual CF is not esential but just a nice bonus all of that.
Jiri
Joe,
Using the darlington based IP stage does have some advantages but running the 'front' pair of BJT's at around 30uA as you have, biased with 22k R's is going to degrade the noise performance significantly.
Need to run more current to achieve noise specs these BJT's are capable of. As always it is a compromise, in this case between IP bias current and noise spec.
In my own journey looking for ULN 'super beta' BJT's, one thing that did pique my interest is the package dependent beta variation of SSM2212. SOIC package typical Beta is 605 @ 1mA. LFCSP package typical Beta @ 1mA is 1800.
I have no idea why this is the case (no pun) all other specs are identical. Might be worth testing a few of these parts.
Terry
Terry,
Honestly said I'm a bit surprised to read this when you say that the 22k resistor will result in only 30uA current. Joe is talking about 4 - 5mA current and up to 9 - 10mA depending on the Anode resistors under #810. Your remark would imply that the 22k resistor needs to be lowered significantly in order to receive a much higher current or not? And Joe is even suggesting to leave the 22k resistor out and that would be the equivalent of having a resistor with infinite high resistance at that particular position so quite the opposite. So I'm wondering what is right and what is wrong here.
Could you comment on that? Thanks
Günter
Honestly said I'm a bit surprised to read this when you say that the 22k resistor will result in only 30uA current. Joe is talking about 4 - 5mA current and up to 9 - 10mA depending on the Anode resistors under #810. Your remark would imply that the 22k resistor needs to be lowered significantly in order to receive a much higher current or not? And Joe is even suggesting to leave the 22k resistor out and that would be the equivalent of having a resistor with infinite high resistance at that particular position so quite the opposite. So I'm wondering what is right and what is wrong here.
Could you comment on that? Thanks
Günter
Hi Terry
I hear you, but...
I seem to recall having this discussion before.
The stage I posted in #803 and here is the part where the Darlington pair sits:
As we both know, I do not use 2SC2547 pair. They are different, as per our dicussion.
But there was something that emerged afterwards that we never discussed. That 22K limits maximum voltage swing in that particular application, for example if this was a power amp and not a preamp. If you don't need to swing a lot of volts coming off the Anode of the 6922, then I would use 1K. If this was used in a power amp, then you cannot use anything that low.
In the power amp I have here, that needs to swing 200V, then no resistor can be had in that circuit. The output is current and that is current driving the 6922 through the Cathode. The Darlington is a transconductance stage, volts converted to current.
That resistor, depending on value, analyse the circuit and it will limit the final voltage swing the stage can achieve.
Like you said, there is a compromise(s) to be made.
Cheers, Joe
Terry is talking about the current of the first transistor driving the base of the second being 25-30uA with 22K resistor (0.6V/22K=0.000027A=27uA).
Adding a lower value resistor there increase that current. But in the application here, there is a downside, it has an effect on the max voltage swing available to the stage. I only found this out in practice when I ran into trouble. In a power amp where the output tubes have near to no gain (1.5x, which is very unusual) and the need to swing crazy amounts of voltage, then you can't use a resistor at all. But another way to improve the situation was found, but I was helped by Terry, so credit to him.