This is a funny way of using a cascode as phono-amp. I used it in the 80s, came to thinka about when discussing cascodes with SY.
Originally it was with a cascoded 12AX7, with in my eyes to high ressitance values and also a ECC82 White follower. So I added the extra current feed to the lower cascode triode and simplified the CF.
Originally it was with a cascoded 12AX7, with in my eyes to high ressitance values and also a ECC82 White follower. So I added the extra current feed to the lower cascode triode and simplified the CF.
An externally hosted image should be here but it was not working when we last tested it.
Lars,
What is funny about it?
First stage has Zout tied down reasonably tight, suitable for driving the RIAA equalisation, although I would run lower RL than the 120K (R5) you show and up the current through the cascode. That may well make R4 redundant. Then adjust the RIAA components as required.
That is as long as you can get enough gain out of the stage.
(Hint for others - gain of a cascode is approximately the RL on the top triode X the gm of the lower triode, Thats why R5 is higher than you might expect and why R4 has been added to increase the current in the lower triode and hence get the gm up a bit).
I'm not that fussed about the RIAA components going back to rail like that rather than down to 0V. A good demonstrator to newbies about what we mean when we say that the high voltage rail IS AC ground.
The output circuit would be a really good test of one theory/opinion that is floating around. That theory is that the cathode follower is analagous to a common collector transistor circuit and that you need (for in depth analysis) to consider its input as being between grid and anode rather than between grid and cathode. The "required" current source load for the cathode follower is there.
I should think that it would work very well.
A bit of lateral thinking in there - I'm in favour of that.
So if you were expecting "flames" you may still get them but not from me.
Cheers,
Ian
What is funny about it?
First stage has Zout tied down reasonably tight, suitable for driving the RIAA equalisation, although I would run lower RL than the 120K (R5) you show and up the current through the cascode. That may well make R4 redundant. Then adjust the RIAA components as required.
That is as long as you can get enough gain out of the stage.
(Hint for others - gain of a cascode is approximately the RL on the top triode X the gm of the lower triode, Thats why R5 is higher than you might expect and why R4 has been added to increase the current in the lower triode and hence get the gm up a bit).
I'm not that fussed about the RIAA components going back to rail like that rather than down to 0V. A good demonstrator to newbies about what we mean when we say that the high voltage rail IS AC ground.
The output circuit would be a really good test of one theory/opinion that is floating around. That theory is that the cathode follower is analagous to a common collector transistor circuit and that you need (for in depth analysis) to consider its input as being between grid and anode rather than between grid and cathode. The "required" current source load for the cathode follower is there.
I should think that it would work very well.
A bit of lateral thinking in there - I'm in favour of that.
So if you were expecting "flames" you may still get them but not from me.
Cheers,
Ian
Another thing is that in practice, beyond the simulator, the passive RIAA components may need readjustment. It seems that most tube models aren't that real with Ri especially when cascoding them. Had to realign the Steve Bench 12AY7 paralleled with ECC88 cascode on top. A makeshift Hagerman inverse Riaa will prove most practical to feed an FFT impulse to the phono input, so to check for a flat output.
Cascode makes awesome input stages 
http://geek.scorpiorising.ca/GeeK_ZonE/index.php?topic=2879.msg33448#msg33448
Cheers!
http://geek.scorpiorising.ca/GeeK_ZonE/index.php?topic=2879.msg33448#msg33448
Cheers!
Some tend to sound weird in my experience. Especially if they repeat cascoding stage after stage. They tend to highten detail and lose flow. That is subjective experience of course, you may pass it. For input is good to cascode (bandwidth) or even essential, to keep Miller low (MM carts). Best I ever experienced is Steve Bench circuit. For other schematics that look good or better on paper, can't never be sure of the final result. Iron must burn, solder flow, vinyl spin.
P.S. Many fine RIAA circuits about, those two here look expertly nice too. What we miss here in the tube forum is a clever and rather simple current sourced shunt regulator. Adjustable from 200-400V. Especially for RIAA, the PSU is more than 50% of the sound.
The solid state guys have put a big group buy for hundreds of Toolreg boards. They have heard the niceness. They have made the simple good thing, they roll. We must develop one too. Maybe Mosfet controlled by tube. Any initial circuit ideas?
P.S. Many fine RIAA circuits about, those two here look expertly nice too. What we miss here in the tube forum is a clever and rather simple current sourced shunt regulator. Adjustable from 200-400V. Especially for RIAA, the PSU is more than 50% of the sound.
The solid state guys have put a big group buy for hundreds of Toolreg boards. They have heard the niceness. They have made the simple good thing, they roll. We must develop one too. Maybe Mosfet controlled by tube. Any initial circuit ideas?
Ian,
With "funny" I ment: "Unusual" with a, not so common, little twist. Also remember I have used and played this stage both with ECC83/ECC82, 6CW4 and E88CC. This was to show an interesting circuit from the 80´s left in the closet that sounded OK in those days! It will still remain there as it isn´t so "funny" anymore.
And Ian, why should I expect flames from you, you haven´t got any from me either!
The reason for publishing it where to discuss it as the subject of cascodes where brought up in another thread where SY participated.
About the RIAA going to rail it is the natural choice here as gain is calculated Ra*Gm and Ra is frequencydependant altered.
SY,
Didn´t expect any ovations and didn´t want any though you seem to like cascodes. You are also aware I am not in favor of them! I used unbypassed cathode for PU with higher output. I used it with bypass caps for lower outputs as diodes wheren´t in fasion at that time. Gain is 6dB higher, 35dB, with diode or bypassed. You have to be careful with choosen gain as overload reserve is small.
Salas,
The inverse RIAA at the the input is based on Hagermans model right out of the box. By the way Hagermans ideas aren´totally correct about the fourth timeconstant as Neumann uses a second order filter at 50kHz. As I have the RIAA built up in three versions now standing in the closet I know this one is close to correct. Haven´t heard of RIAAs that didn´t need trimming. So feeding the input as you say is not more than theoretical correct. A RIAA must be trimmed in combination with PU and player from recorded frequency tracks. I have an old DECCA, maybe there are better ones out there nowadays.
Good idea about PSU, though. This one is fed by a sand, regulated PSU from Erik A.(Audion).
Guys,
From a theoretical view you could call the cacode "cascade" as the cascode could be looked upon as a DC-coupled cascaded GC into a GG!
With "funny" I ment: "Unusual" with a, not so common, little twist. Also remember I have used and played this stage both with ECC83/ECC82, 6CW4 and E88CC. This was to show an interesting circuit from the 80´s left in the closet that sounded OK in those days! It will still remain there as it isn´t so "funny" anymore.
And Ian, why should I expect flames from you, you haven´t got any from me either!
The reason for publishing it where to discuss it as the subject of cascodes where brought up in another thread where SY participated.
About the RIAA going to rail it is the natural choice here as gain is calculated Ra*Gm and Ra is frequencydependant altered.
SY,
Didn´t expect any ovations and didn´t want any though you seem to like cascodes. You are also aware I am not in favor of them! I used unbypassed cathode for PU with higher output. I used it with bypass caps for lower outputs as diodes wheren´t in fasion at that time. Gain is 6dB higher, 35dB, with diode or bypassed. You have to be careful with choosen gain as overload reserve is small.
Salas,
The inverse RIAA at the the input is based on Hagermans model right out of the box. By the way Hagermans ideas aren´totally correct about the fourth timeconstant as Neumann uses a second order filter at 50kHz. As I have the RIAA built up in three versions now standing in the closet I know this one is close to correct. Haven´t heard of RIAAs that didn´t need trimming. So feeding the input as you say is not more than theoretical correct. A RIAA must be trimmed in combination with PU and player from recorded frequency tracks. I have an old DECCA, maybe there are better ones out there nowadays.
Good idea about PSU, though. This one is fed by a sand, regulated PSU from Erik A.(Audion).
Guys,
From a theoretical view you could call the cacode "cascade" as the cascode could be looked upon as a DC-coupled cascaded GC into a GG!
I avoid the ''lost'' constant too. It peaks ultrasonically, disrupts circuits, does nothing good. I have designed a jfet Riaa, had checked it with Telarc Omnidisc and FFT, and tuned the Bench in comparison. The Hagerman, I meant to really build it and feed FFT. Not only to help modeling. Its alternative to PU. And yes, the shunt PSU we must make. What HV Mosfet you would pick for CCS and main shunt element?
Hi Salas,
About the PSU I really don´t know, have not been into that yet. The old one I have is all BJT and not shunt.
Read somewhere about the RIAA, that if you choose 100kHz as 4th constant you should get the best phase response with Neumann engraved records. It seems like there is no "perfect" RIAA IRL!
About the PSU I really don´t know, have not been into that yet. The old one I have is all BJT and not shunt.
Read somewhere about the RIAA, that if you choose 100kHz as 4th constant you should get the best phase response with Neumann engraved records. It seems like there is no "perfect" RIAA IRL!
I simulated and made one for the jfet riaa of mine, a led referenced mosfet ccs feeding a no loop feedback mosfet shunt, it really made that phono show its teeth. I tell you we must come up with something for the tubes too. It is beyond perfecting your already excellent circuits. Nitro stuff. I can send you my low voltage one and see if you can start from that for HV.
I hold up my hand as the originator of the "4th T/C' in RIAA networks. Jim Hagerman got it from me. During the late 70's, my Guru Rown McCombe was finding that things sounded better with that ever decreasing 75uS roll off flattened out, and by ear he found it to SOUND best around 50kHz.
We then called around the Australian agents for cutting gear to ask, no one knew anything except the Neiman tech who gave us a figure of 3.18uS which is pretty dang close to 50 kHz. He never mentioned 12dB/oct though.
I published this in my "TubePreamp CookBook" of 1995, and quite a few people have picked it up since.
But Rowan never suggested using a fixed 3/18uS point, all his phono stages had a R trimmer to set the 75uS point, and an air C trimmer to set the '3.18" point, all to be done off (music) discs by ear. I make no claim to be able to hear adjustments on a 50pF trimmer in an RIAA network, but he could, repeatablly, and the end results were always superb, sonically!
Regards, Allen (Vacuum State)
We then called around the Australian agents for cutting gear to ask, no one knew anything except the Neiman tech who gave us a figure of 3.18uS which is pretty dang close to 50 kHz. He never mentioned 12dB/oct though.
I published this in my "TubePreamp CookBook" of 1995, and quite a few people have picked it up since.
But Rowan never suggested using a fixed 3/18uS point, all his phono stages had a R trimmer to set the 75uS point, and an air C trimmer to set the '3.18" point, all to be done off (music) discs by ear. I make no claim to be able to hear adjustments on a 50pF trimmer in an RIAA network, but he could, repeatablly, and the end results were always superb, sonically!
Regards, Allen (Vacuum State)
Salas is asking about a group buy on a tube voltage current sourced shunt reg.
I (again) claim major publication of this concept, also in my 1995 TubePreamp CookBook, and since then must have sold at leat 500 kits to build what I call the "SuperReglator".
After many attemps to do it with tubes, I gave up because the DC points would always drift all over the place, and did it with semis.
MOSFETs were found to be far more reliable than bipolars, so it ended up with a MOSFET CCS feeding a opamp (LF351 or AD797) /MOSFET shunt. Most people like the sonic results, and we now have a new version under test that uses a AD811 opamp in a different topology.
Happy to workout a group buy price for this crowd if there is an interest.
Regards, Allen (Vacuum State)
I (again) claim major publication of this concept, also in my 1995 TubePreamp CookBook, and since then must have sold at leat 500 kits to build what I call the "SuperReglator".
After many attemps to do it with tubes, I gave up because the DC points would always drift all over the place, and did it with semis.
MOSFETs were found to be far more reliable than bipolars, so it ended up with a MOSFET CCS feeding a opamp (LF351 or AD797) /MOSFET shunt. Most people like the sonic results, and we now have a new version under test that uses a AD811 opamp in a different topology.
Happy to workout a group buy price for this crowd if there is an interest.
Regards, Allen (Vacuum State)
I use that 3.18us trick and think it's worthwhile. I doubt that my 5.4 decade old ears could distinguish 3 from 3.18, but the presence of it does seem to help realism in a subtle way.
Lars, your comments about overload are very well taken. I think is a good thing to get as much gain out of the first stage as possible, so the use of all-in-one passive EQ helps overload with minimal compromise on noise. My own phono stage that I used from about 1980 until about two years ago used a FET-tube cascode in that position, not unlike Allen's designs. That allowed a good deal of gain upfront and the result was a very quiet unit.
Lars, your comments about overload are very well taken. I think is a good thing to get as much gain out of the first stage as possible, so the use of all-in-one passive EQ helps overload with minimal compromise on noise. My own phono stage that I used from about 1980 until about two years ago used a FET-tube cascode in that position, not unlike Allen's designs. That allowed a good deal of gain upfront and the result was a very quiet unit.
Allen Wright said:
Hi Allen,
Thanks for your reminders here and your significant contribution to the hobby.
To clarify, I was not asking for a group buy of something certain yet. I was trying to make a point about designing a simplest possible variable Mosfet shunt reg here as a community. I have created a couple for JFET phono circuits of mine and I know their strong sonic boost benefits. My opinion is that we lack an analogous one for the tube line and phono stages in this forum section. So a good example are the solid state guys next door that singled out the easy and good Tool one (DIY A member design), named it Toolreg, tested it, then designed an excellent PCB, and now they run a major group buy for ordering them PCBs. A DIY A community effort. I like that spirit.
On the other hand, if my proposition falls short of interest, people here all well reminded that they can rely on your proven and strong performing design, even better for a benefiting group buy.
I have also used the 3.18u as mentioned the last years . This is done by the added R7 in the schematic. It wasn´t used when I originally built the RIAAs.
Alllen, I wasn´t aware of the 12dBs, and don´t even know if it is correct, until some week ago when I googled and read this, where you also get credit:
"This is done to compensate for the frequency response of real-world cutting machines, which can't keep boosting treble out to infinity and hence have to roll off somewhere above 20kHz, violating the RIAA 'standard' in the process. It seems wise to compensate for this deviation during replay, as Vacuum State (Allen Wright), Graham Slee, Whest Audio, and an army of DIYers found a long time ago. The silly thing now is that often the proposed replay compensation entails no more than a single zero at 50kHz, while e.g. Neumann's cutting curve happens to be treble-limited with a second-order Sallen-Key filter (-3dB at 33kHz in the old SX-66, 50kHz in the ubiquitous SX-74 cutter). Replay with a single-order zero at 50kHz then redresses the phase response somewhat, but leads to excess treble between 10 and 20kHz!"
"Have a look at the (above) figures, which are simulations of a Neumann SAL-74B cutter amplifier followed with standard RIAA de-emphasis (red) and single-zero-50kHz-compensated de-emphasis (blue). The standard equalisation has a near-flat amplitude response out to 20kHz, but suffers over 25 degrees of phase shift at this frequency. The 'compensated' replay equalisation shows a slowly rising treble, but has less phase shift. (The green curves are for a single-zero 100kHz compensation, something AFAIK no-one uses.)
In the end neither solution is right or wrong, although it must be remarked that the treble peaking in the compensated version might offset some of the treble losses inherent to vinyl's crude manufacturing process. Nevertheless, the correct replay of LPs remains an illusive target. "
Please give your comments on this as it might, as I said earlier, not be correct.
Must also check the theoretical difference between compensated and not when I get back to the computer I used when drawing the schematic.
Alllen, I wasn´t aware of the 12dBs, and don´t even know if it is correct, until some week ago when I googled and read this, where you also get credit:
"This is done to compensate for the frequency response of real-world cutting machines, which can't keep boosting treble out to infinity and hence have to roll off somewhere above 20kHz, violating the RIAA 'standard' in the process. It seems wise to compensate for this deviation during replay, as Vacuum State (Allen Wright), Graham Slee, Whest Audio, and an army of DIYers found a long time ago. The silly thing now is that often the proposed replay compensation entails no more than a single zero at 50kHz, while e.g. Neumann's cutting curve happens to be treble-limited with a second-order Sallen-Key filter (-3dB at 33kHz in the old SX-66, 50kHz in the ubiquitous SX-74 cutter). Replay with a single-order zero at 50kHz then redresses the phase response somewhat, but leads to excess treble between 10 and 20kHz!"
"Have a look at the (above) figures, which are simulations of a Neumann SAL-74B cutter amplifier followed with standard RIAA de-emphasis (red) and single-zero-50kHz-compensated de-emphasis (blue). The standard equalisation has a near-flat amplitude response out to 20kHz, but suffers over 25 degrees of phase shift at this frequency. The 'compensated' replay equalisation shows a slowly rising treble, but has less phase shift. (The green curves are for a single-zero 100kHz compensation, something AFAIK no-one uses.)
In the end neither solution is right or wrong, although it must be remarked that the treble peaking in the compensated version might offset some of the treble losses inherent to vinyl's crude manufacturing process. Nevertheless, the correct replay of LPs remains an illusive target. "
Please give your comments on this as it might, as I said earlier, not be correct.
Must also check the theoretical difference between compensated and not when I get back to the computer I used when drawing the schematic.
Here is what happened to my RIAA without and with the 3.18u resistor. +6.9dB at 100kHz. Member jackinnj had seen potential ultrasonic problems when simulating my circuit for crosschecking. The labels above the plots are misplaced. The skyrocketing one is with 3.18u provision.
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.