Steven,
Glad my reply was helpful. Great idea to post pictures as you make progress. I found gabo’s photos published earlier in this thread very helpful.
Perhaps we could revive the Wiki that existed for the first version of the Baby Huey build and add construction of the newer PCB’s and power supply board.
Francois.
Hi,
based on new page on Merlin's site:
The Valve Wizard
It is easy to develop a system to autobias this amp at every startup (or on demand). Who's interested into implementing it on the project?
I will post this comment on both EL84 and EL34 threads.
based on new page on Merlin's site:
The Valve Wizard
It is easy to develop a system to autobias this amp at every startup (or on demand). Who's interested into implementing it on the project?
I will post this comment on both EL84 and EL34 threads.
There is a proven design on a dutch diyaudio site. Have not implemented it yet. But gerbers etc are available in the rars
Bias automaat - forum.zelfbouwaudio.nl
Bias automaat - forum.zelfbouwaudio.nl
You are correct. The original BH PCB was developed to function without an of-board HV supply. But do note that the board was also developed with the assumption that you will provide a DC supply for the heaters, and an AC supply for bias. A good review of the board would be advised, and look at the earlier posts in this thread by Marc (board designer) about wiring the boards.
Hi Myles
Please see attached PDF, it might not be the latest revision but there are some guidelines.
BR
Eric
Hi Eric et al,
Thank you for the help. I have this pdf already. I was going to keep my more recent BHEL34 pcb, but a member replied to my advertisement and I will let him have it to cpmplete his buy, and I will keep the original (like in the pdf). I have a lot of parts for it already.
Regards,
MM
Thank you for the help. I have this pdf already. I was going to keep my more recent BHEL34 pcb, but a member replied to my advertisement and I will let him have it to cpmplete his buy, and I will keep the original (like in the pdf). I have a lot of parts for it already.
Regards,
MM
Finished cleaning the workshop and started to load the Marc/Prasi boards using the Mouser supplied parts.
First note:
the 220K R7 and R8 should be 2W not the 0.6W which is specified on the Mouser List. This is for its Voltage rating NOT for its power rating. Particularly important for guys building KT88 etc versions with higher rails. I had some 240K 2W "specials" in my parts bins and used them instead.
Have'nt loaded R28/D5 etc for the negative rail as the power tranny I am using has a 40V bias winding and it will not give me a high enough -ve supply.
PLITRON 754709 with 340V @ 0.7 Amps, 40V @ 0.1 Amps and 6.3V @ 6.8 Amps.
The -ve rail needs to be approx. 3 times the output tube bias voltage to make sure the MOSFET source followers never go into cutoff. I worry about that with some of your KT88 builds. Do you have high enough negative rail?
I will be voltage doubling the 40V winding (external to the PCB) to make sure I have plenty of -ve rail.
Just some other notes for High B+ builders.
You can adjust R16,R17 to keep LED1 current at around 2mA - Not critical
You can adjust R33, R39 to increase the current in the source follower for better driving the higher required output tube grid drive voltages (actually to drive the grid capacitance to higher voltage at high frequencies, preventing slewrate limiting) - watch the mosfet dissipation and add a clip on heatsink if required. Remember that dissipated power is a current squared thing.
Tossing up wether to fit the Hammond 1650P (Raa = 6K6) or the Hammond 1650N (Ra = 4K3). I have a pair of each. Still planning on KT77 outputs.
Cheers,
Ian
First note:
the 220K R7 and R8 should be 2W not the 0.6W which is specified on the Mouser List. This is for its Voltage rating NOT for its power rating. Particularly important for guys building KT88 etc versions with higher rails. I had some 240K 2W "specials" in my parts bins and used them instead.
Have'nt loaded R28/D5 etc for the negative rail as the power tranny I am using has a 40V bias winding and it will not give me a high enough -ve supply.
PLITRON 754709 with 340V @ 0.7 Amps, 40V @ 0.1 Amps and 6.3V @ 6.8 Amps.
The -ve rail needs to be approx. 3 times the output tube bias voltage to make sure the MOSFET source followers never go into cutoff. I worry about that with some of your KT88 builds. Do you have high enough negative rail?
I will be voltage doubling the 40V winding (external to the PCB) to make sure I have plenty of -ve rail.
Just some other notes for High B+ builders.
You can adjust R16,R17 to keep LED1 current at around 2mA - Not critical
You can adjust R33, R39 to increase the current in the source follower for better driving the higher required output tube grid drive voltages (actually to drive the grid capacitance to higher voltage at high frequencies, preventing slewrate limiting) - watch the mosfet dissipation and add a clip on heatsink if required. Remember that dissipated power is a current squared thing.
Tossing up wether to fit the Hammond 1650P (Raa = 6K6) or the Hammond 1650N (Ra = 4K3). I have a pair of each. Still planning on KT77 outputs.
Cheers,
Ian
Last edited:
Roberto,
With a bit of "cut and paste" itt should work.
Best:
Duplicate the MOSFET Source followers (and bias feed) and the output tubes.
2nd best:
If just paralleling output tubes then the 2 on each side would need to be matched.
Remember that the BH scheme relies on forced balance and also trading output tube gm for reduced rp. Parallel output tubes act like a single "composite" tube with 2x the gm and 1/2 the rp. Should work fine, in fact you may be able to reduce R13 which sets the shunt feedback level.
I just noted that R13 is shown as 39K on the schematic. For those guys who have their amps built and running I strongly suggest that you try lower values for R13, For the EL84 build I ended up with 16K. At 39K it sounded a bit solidstate'ish. I tried everything 10K to 47K and there was a definite optimum at around 16K to 18K. The lower values mean less correction of the output tubes but also less stress on the diffamp.
Think of R13 as your amp "tune" resistor.
If you want to do a "quick and dirty" trial just tack another 39K across the existing 39K (to give 19K5) and see if you like it any better
Cheers,
Ian
Cheers,
Ian
With a bit of "cut and paste" itt should work.
Best:
Duplicate the MOSFET Source followers (and bias feed) and the output tubes.
2nd best:
If just paralleling output tubes then the 2 on each side would need to be matched.
Remember that the BH scheme relies on forced balance and also trading output tube gm for reduced rp. Parallel output tubes act like a single "composite" tube with 2x the gm and 1/2 the rp. Should work fine, in fact you may be able to reduce R13 which sets the shunt feedback level.
I just noted that R13 is shown as 39K on the schematic. For those guys who have their amps built and running I strongly suggest that you try lower values for R13, For the EL84 build I ended up with 16K. At 39K it sounded a bit solidstate'ish. I tried everything 10K to 47K and there was a definite optimum at around 16K to 18K. The lower values mean less correction of the output tubes but also less stress on the diffamp.
Think of R13 as your amp "tune" resistor.
If you want to do a "quick and dirty" trial just tack another 39K across the existing 39K (to give 19K5) and see if you like it any better
Cheers,
Ian
Cheers,
Ian
Last edited:
Roberto,
It certainly could but my reservations would be in the physical implementation.
R13 and any added pot are at high voltage. You need to keep their connections well away from any 0V lines. Using a high voltage rated pot would mean a panel mount job and then the wiring capacitance is going to screw with the feedback at higher frequencies so a panel mount pot is not recommended.
This is exactly what I did in my set to work/tune. The trimmer was a good Bourns 10 turn reliable job and the R13 + trimmer sat up in the air in a mechanically unsound but low stray capacitance arrangement. Once I found what value I needed, a properly secured fixed resistor was substituted.
This area around R13 has the most affect on sound and it is important to keep stray capacitance low. Putting a pot at the end of some hook up wires may sound different.
Believe it or not I even noticed the sonic difference with resistors of the same value from different manufacturers.
This is expected, what ever you put in the feedback path reflects into the amps response.
ASIDE: I have actually seen some SS guys make use of this by using Riken Ohm Carbon Composition resistors in the feedback path to deliberately put in 2nd harmonic distortion, to help mask odd order distortions, to make the amp sound better. Of course it also added noise and was a case of fixing the symptom and not the disease.
Just assume, that to a fair extent, you get the inverse of the feedback network response as the amp response, so keep wiring around R13 minimalist and neat and tidy.
Cheers,
Ian
It certainly could but my reservations would be in the physical implementation.
R13 and any added pot are at high voltage. You need to keep their connections well away from any 0V lines. Using a high voltage rated pot would mean a panel mount job and then the wiring capacitance is going to screw with the feedback at higher frequencies so a panel mount pot is not recommended.
This is exactly what I did in my set to work/tune. The trimmer was a good Bourns 10 turn reliable job and the R13 + trimmer sat up in the air in a mechanically unsound but low stray capacitance arrangement. Once I found what value I needed, a properly secured fixed resistor was substituted.
This area around R13 has the most affect on sound and it is important to keep stray capacitance low. Putting a pot at the end of some hook up wires may sound different.
Believe it or not I even noticed the sonic difference with resistors of the same value from different manufacturers.
This is expected, what ever you put in the feedback path reflects into the amps response.
ASIDE: I have actually seen some SS guys make use of this by using Riken Ohm Carbon Composition resistors in the feedback path to deliberately put in 2nd harmonic distortion, to help mask odd order distortions, to make the amp sound better. Of course it also added noise and was a case of fixing the symptom and not the disease.
Just assume, that to a fair extent, you get the inverse of the feedback network response as the amp response, so keep wiring around R13 minimalist and neat and tidy.
Cheers,
Ian
Last edited:
Hey Ian, I was looking back at some of your notes. Shouldn't this be R33 and R38 instead of R39?
I also currently have R7/R8 as the 0.6w versions and have been running the amp for a month or so. Is it worth the preventative to swap those out with 2w versions or just leave it?
Also, just FYI, I'll try playing around with R13 a bit. However, I have the 39K in mine and it sounds great. Of course I come from a background of SS amps, so maybe I don't know what I'm missing
But I do have it apart right now working on the chassis, so easy enough to throw another one across there to see how it sounds. gabo
Last edited:
R33 and R38 is right - my bad.
R7/R8 - if them amp is happy with these at the 0.6W jobs (350V rated) then just leave them in. If building from new I would however change them to 500V rated (2W parts).
Some issues yesterday - that Plitron power tranny is not going to suit my build (bummer since I have 6 of them), just too high a HV winding which would probably give me around +450V HT. Looking at other options from my shelf.
Ian
Edit: Transformer will be OK after all. Just reviewed the Power Supply board schematic. Even has the voltage doubler I need for -ve rail.
Will need to change some 400V rated caps on the amp boards.
R7/R8 - if them amp is happy with these at the 0.6W jobs (350V rated) then just leave them in. If building from new I would however change them to 500V rated (2W parts).
Some issues yesterday - that Plitron power tranny is not going to suit my build (bummer since I have 6 of them), just too high a HV winding which would probably give me around +450V HT. Looking at other options from my shelf.
Ian
Edit: Transformer will be OK after all. Just reviewed the Power Supply board schematic. Even has the voltage doubler I need for -ve rail.
Will need to change some 400V rated caps on the amp boards.
Last edited:
Fitted all the passive components. LEDs and diodes to PS Board today. Stopped work while sort out the heatsinks etc.
Also I am not understanding the IRF9610 circuit.
I had assumed it was a simple capacitance multiplier but no.
power supply - Mosfet based ripple filter for tube amp - Electrical Engineering Stack Exchange
That example is for a +ve supply not a negative supply so for -ve supply you use a P Channel Mosfet (so the IRF9610 is correct) and you turn around the protection diode D8 (also correct).
BUT the Mosfet on the board has the Drain and Source swapped compared to what is required.
I think what is happening on the BH PS board is the intrinsic MOSFET bulk diode (from Drain to Source with anode at Drain) is just conducting and you are getting no capacitance multiplier affect at all. Any filtering is from R1 and R6 feeding C6 via the Mosfet bulk diode. For a "real" capacitance multiplier R1 is not needed and R6 is way to high.
Is this a design oversight or am I missing something?
Also D8 and D11 need to be mounted on the back of the PCB else they crash into the Heatsinks supplied by Mouser.
Cheers,
Ian
Also I am not understanding the IRF9610 circuit.
I had assumed it was a simple capacitance multiplier but no.
power supply - Mosfet based ripple filter for tube amp - Electrical Engineering Stack Exchange
That example is for a +ve supply not a negative supply so for -ve supply you use a P Channel Mosfet (so the IRF9610 is correct) and you turn around the protection diode D8 (also correct).
BUT the Mosfet on the board has the Drain and Source swapped compared to what is required.
I think what is happening on the BH PS board is the intrinsic MOSFET bulk diode (from Drain to Source with anode at Drain) is just conducting and you are getting no capacitance multiplier affect at all. Any filtering is from R1 and R6 feeding C6 via the Mosfet bulk diode. For a "real" capacitance multiplier R1 is not needed and R6 is way to high.
Is this a design oversight or am I missing something?
Also D8 and D11 need to be mounted on the back of the PCB else they crash into the Heatsinks supplied by Mouser.
Cheers,
Ian
Last edited:
Sorry to pepper you guys but 3hrs down the track with luddite pencil and paper design, I am convinced we have a problem.
The source and Drain connections to that IRF9610 are REVERSED.
The power supply is working to give you full voltage and full ripple because the bulk diode which is intrinsic to the MOSFET is forward biased and is effectively taking the MOSFET out of circuit.
Also several component values are suboptimum or worse.
What do we need to fix it?
First we need to isolate the MOSFET Drain and Source connections:
So
Do not fit (or remove) R1, R6 and D8
Next make a new Drain connection:
Wire from R6 pad next to the big 2200uF/25V cap to the R1 pad which is closest to the board edge. Note: rather than a wire link you can use a low value resistor, say 47 Ohms which acts as a Drain Stop resistor against oscillation,. Rarely required, just use a piece of wire
Next make a new Source Connection:
Wire from D8 Cathode Pad to The R6 pad which joins C6. I strongly recommend that in this case you use a 100 Ohm resistor to make this link (see below).
WE need to restore the protection diode, the old anode connection is fine but the cathode needs to connect to that R6 pad which joins C6 noted above. Why not to the Source directly? This is a Gingertube "smart ****" mod, by connection the output side of the 100 Ohm mentioned above we add current limiting protection to the supply. Since Mosfet Gate/Source Thresholds are all over the place (IRF9610 says min 2V max 4V) that will introduce a 20 to 40mA current limit if the output gets shorted. AOT (Adjust on Test which is the shorthand form of the "design engineer saying he really doesn't know").
Next R2 is way too high. The circuit operates by limiting output voltage to just less than the lowest dip in the input voltage. So R2 should drop a DC voltage of at least the maximum peak to peak ripple voltage from the raw supply. This is a lowish current supply and ripple is unlikely to be huge so I would try 22K for R2. A higher value means more immunity to ripple but also a lower output voltage and more dissipation in the MOSFET.
I will actually be building my board this way and will report when running.
Would be glad to hear someone say I think your WRONG but this is my current thinking.
Cheers
Ian
Edit/PS How important is a low ripple -ve supply. For the BH its is far less important than usual. The current source loads on the Source Followers isolate -ve rail noise and the bias feeds have additional dropping resistors and filtering. So you may consider a 90% fix as just leaving everything as is, or removing the mosfet and putting a diode from Drain to Source, which is what you have now.
But to be OCD and have the cleanest B- and bias supplies the mod should do it.
The source and Drain connections to that IRF9610 are REVERSED.
The power supply is working to give you full voltage and full ripple because the bulk diode which is intrinsic to the MOSFET is forward biased and is effectively taking the MOSFET out of circuit.
Also several component values are suboptimum or worse.
What do we need to fix it?
First we need to isolate the MOSFET Drain and Source connections:
So
Do not fit (or remove) R1, R6 and D8
Next make a new Drain connection:
Wire from R6 pad next to the big 2200uF/25V cap to the R1 pad which is closest to the board edge. Note: rather than a wire link you can use a low value resistor, say 47 Ohms which acts as a Drain Stop resistor against oscillation,. Rarely required, just use a piece of wire
Next make a new Source Connection:
Wire from D8 Cathode Pad to The R6 pad which joins C6. I strongly recommend that in this case you use a 100 Ohm resistor to make this link (see below).
WE need to restore the protection diode, the old anode connection is fine but the cathode needs to connect to that R6 pad which joins C6 noted above. Why not to the Source directly? This is a Gingertube "smart ****" mod, by connection the output side of the 100 Ohm mentioned above we add current limiting protection to the supply. Since Mosfet Gate/Source Thresholds are all over the place (IRF9610 says min 2V max 4V) that will introduce a 20 to 40mA current limit if the output gets shorted. AOT (Adjust on Test which is the shorthand form of the "design engineer saying he really doesn't know").
Next R2 is way too high. The circuit operates by limiting output voltage to just less than the lowest dip in the input voltage. So R2 should drop a DC voltage of at least the maximum peak to peak ripple voltage from the raw supply. This is a lowish current supply and ripple is unlikely to be huge so I would try 22K for R2. A higher value means more immunity to ripple but also a lower output voltage and more dissipation in the MOSFET.
I will actually be building my board this way and will report when running.
Would be glad to hear someone say I think your WRONG but this is my current thinking.
Cheers
Ian
Edit/PS How important is a low ripple -ve supply. For the BH its is far less important than usual. The current source loads on the Source Followers isolate -ve rail noise and the bias feeds have additional dropping resistors and filtering. So you may consider a 90% fix as just leaving everything as is, or removing the mosfet and putting a diode from Drain to Source, which is what you have now.
But to be OCD and have the cleanest B- and bias supplies the mod should do it.
Last edited: