Gyrator loaded Son of ZV9/F3

When I saw ZV9 for the first time it looked to me like a beautiful preamp topology made to an power amp thanks to existence of LU1014.
Few months ago I started to entertain this single ended preamp idea again and not only that the sound was exceptional (I like it so much that I removed BAF 2013 preamp from the case to move this one in although I thought that I'll keep the BAF2013 for a very long time), it showed other great qualities too: great drive/swing capability (Zout less than 50R and ability to drive power buffers like F4), great versatility in terms of gain and power supply used, low and steady consumption (10-12mA per channel), it's stable into reactive loads and the freq./phase response is linear up to 100kHz.

The story about this preamp actually started here but I think it deserves a dedicated thread. The main change from the original ZV9 is that I use gyrator instead of aleph CCS - it sounds better to me but it's easy to test both.

As a power supply I use 32V SMPS taken from dead HP inkjet printer, followed by a capacitance multiplier. The PCB is 74 x 25mm, copper side view. R9 is mounted on the output terminal and C3 (WIMA-MKP10) is beside the PCB due to its size. C2 is 2-5pF, made from two twisted isolated wires cut to length for best square wave response (I did this just because I don't have small value caps in my parts repository at the moment). As shown, the gain is 13dB (4.5V/V) but can be set as ratio of R1/R3 in a wide range (I tested 6-26dB). The clean swing at the output with 30V rail voltage is 24V peak-to peak (8.5V_RMS) into 10k load.

EDIT:
One thing I forgot to mention - due to RC constants present in the preamp and cap. multiplier, 10-15 seconds are needed for capacitors to charge and the sound starts to come out. You might call it: very slow start :D
 

Attachments

  • pre-sch.gif
    pre-sch.gif
    6.8 KB · Views: 4,219
  • psu-sch.gif
    psu-sch.gif
    3.9 KB · Views: 4,151
  • pcb-layout.gif
    pcb-layout.gif
    34.8 KB · Views: 3,935
  • pcb-made.jpg
    pcb-made.jpg
    65.4 KB · Views: 3,715
  • modul.jpg
    modul.jpg
    159.8 KB · Views: 3,657
  • built.jpg
    built.jpg
    195.3 KB · Views: 1,766
Last edited:
Hi Juma,
being me a Juma-follower (no, it's not a new topology :p) I have already collected all the components for this new pre-amp (and drawn the PCB before you published this one...).
A question about the JFETs: I do have many BF862, but they are all in the range >16mA. If I'm well understanding, it's all matters of current dissipated in the BC and in the BS, isn't it? (I've built some copper TO-92 heatsinks)
And (you always replied to a similar question in the other thead), would a 2SK170V be used instead? (with Idss in 12-15mA range).

Many thanks again
Guido
 
... it's all matters of current dissipated in the BC and in the BS, isn't it? (I've built some copper TO-92 heatsinks)
Hi Guido, it's not about current only but power (P = V x I ) but I guess that's what you meant and you got it right. If you keep the power dissipation on these two to less than 200mW the heatsink is not needed. You have to know what is the max. voltage swing you want at the output and that will determine the optimal power supply voltage (which influences the power disipation in BJT and MOSFET) . If you don't need more than 19V_peak-to-peak (6.7V_RMS) at the output, the power supply voltage can be 24V and then you can use 20mA JFET with no heatsinks on BJT and MOSFET (in that case the R22 should be 51k to ensure symmetrical clipping).

... would a 2SK170V be used instead? (with Idss in 12-15mA range)
Sure, but I think it will sound a bit different - compare it with your higher Idss BF862 since you alredy have TO-92 heatsinks to find out what you like more. JFET won't mind higher current since it operates at Vds of about 4V only.
In this preamp I use BF862 with Idss of 14.1 mA and got the Id of 11mA with 4R7 degeneration (R2). If you use, for example, BF862 with Idss of 18 mA I supose you'll get the Id of about 14 mA which is still OK. Do some testing and you'll see... :)
 
Last edited:
Hi Juma,

I would like to try this preamp, but… I can’t find BF862 locally. I have some 2SK170BL (10-11mA Idss) and J310. Can I use any of these, and what is needed to change in this case? I don’t need more than 6V RMS at the output. Thanks.

Zdravko
 
Hi Zdravko,
you can buy BF862 from Farnell.de and Mouser.de online shops but if you are lazy like me, try 11mA k170bl with 24V power supply - only thing to change is R22 = 51k (because of lower PS voltage, in order to ensure symmetrical clipping). Also, you can try different values for R19 (it sets JFET's Vds, between 2-5V you'll find the value that soundwise suites the k170bl the best). k170 has higher parasitic capacitance so you might lower the value of R3 to avoid making the bandwidth narrower i.e. nasty phase shift in the upper range. Of course, that means adjusting the R1 to suit the chosen gain ratio with the new value of R3. C2 might need minor adjustment too. Of course, none of these changes are mandatory, the preamp will work OK as it is, with just k170bl instead of bf862, but I think it won't sound its best...
So, as usual, being lazy means more work :D
 
Last edited:
Superb :D

it's great that power supply is reduced now, however is there any changes for voltage doubler that you previously posted? I already have 2x15V transformer to be used.

this is a request : can I buy from you all transistors used for this project? I'm sure someone else will ask the same
 
...it's great that power supply is reduced now...
It's reduced because I don't need to drive the F4 to full power which is what you asked for.

... any changes for voltage doubler that you previously posted? I already have 2x15V transformer to be used.
You don't need voltage doubler for this version - 30V AC rectified is about 40V DC so drop some voltage with RC filter or use the regulator of your choice.
If you mean to use that transformer with high voltage version from the other thread it's OK, just drop some more voltage on the resistor in the RC filter.


Sorry but I don't have enough parts to sell them. Anyway, they are all current production transistor that can be bought all over the world.
 
When I saw ZV9 for the first time it looked to me like a beautiful preamp topology made to an power amp thanks to existence of LU1014.
Few months ago I started to entertain this single ended preamp idea again and not only that the sound was exceptional (I like it so much that I removed BAF 2013 preamp from the case to move this one in although I thought that I'll keep the BAF2013 for a very long time), it showed other great qualities too: great drive/swing capability (Zout less than 50R and ability to drive power buffers like F4), great versatility in terms of gain and power supply used, low and steady consumption (10-12mA per channel), it's stable into reactive loads and the freq./phase response is linear up to 100kHz.

The story about this preamp actually started here but I think it deserves a dedicated thread. The main change from the original ZV9 is that I use gyrator instead of aleph CCS - it sounds better to me but it's easy to test both.

As a power supply I use 32V SMPS taken from dead HP inkjet printer, followed by a capacitance multiplier. The PCB is 74 x 25mm, copper side view. R9 is mounted on the output terminal and C3 (WIMA-MKP10) is beside the PCB due to its size. C2 is 2-5pF, made from two twisted isolated wires cut to length for best square wave response (I did this just because I don't have small value caps in my parts repository at the moment). As shown, the gain is 13dB (4.5V/V) but can be set as ratio of R1/R3 in a wide range (I tested 6-26dB). The clean swing at the output with 30V rail voltage is 24V peak-to peak (8.5V_RMS) into 10k load.

EDIT:
One thing I forgot to mention - due to RC constants present in the preamp and cap. multiplier, 10-15 seconds are needed for capacitors to charge and the sound starts to come out. You might call it: very slow start :D

Juma, nice and simple as always! :) Couple questions:
1. Whay is R6 referenced to the output?
2. Referencing C8 to the source of jfet (not ground) always sounded better to me, have you tried it?

Cheers, Borko. :)
 
Hi Borko,
1. Whay is R6 referenced to the output?
That's the trick with "modulated cascode" as we now it from ZV9/F3 - practically it's the same idea that is more often used in source follower circuits where the botom end of cascode is modulated from the JFET's source which is the output node. It audibly linearises the operation - I noticed that with it this circuit can be used without the local feedback and it still sounds very good (although the gain is higher) which is not exactly the case with standard cascode.


2. Referencing C8 to the source of jfet (not ground) always sounded better to me, have you tried it?
Connecting the C8 to GND leaves some small "room" (in terms of voltage) for JFET to "breathe" (the Vds is not strictly fixed) and that solution sounds better to me. So, as always, it's matter of personal taste, although in this case the swing on the source of the JFET is very small - it swings only the amount of current that's drawn by the load times 4R7...

Best regards !
:cheers:
 
Hi Dennis,
10k is an optimal value. If you happen to have an extra impotent source component that can not drive a 10k pot, use 25k pot, but in that case the R3 should be changed to 10k and R1 to 47k. If 25k is still too low for the puny source component, add a buffer between it and the pot.
 
A fellow member asked me if this preamp can be made with simple BJTs only so I gave it a try and it works lovely.
He is OK with max. 6V_RMS at the output so the rail voltage is down to 24V.
Since the use of a BJT at the input and single rail PSU imply the use of an input cap I introduced local Schade feedback (R10, R3) to control the gain (about 13dB). Z_out is a bit higher (150-200 R) , Z_in too (47k), some elements' values are changed, but it still sounds very good...
 

Attachments

  • cpsfg.gif
    cpsfg.gif
    5.8 KB · Views: 1,322