Help modifiying schematic

[zafira1981]

Well I'm going crazy trying to modify this schematic for my power supply.

Here is the schem

An externally hosted image should be here but it was not working when we last tested it.

My PSU is

+520DC at +B and +260DC at the center-tap of the power supply caps.

I'm using that PSU because my output tubes are Kt88 and 70W RMS output power instead of 30W.

So I cant use the R's of the PSU original schematic because of the different transformer.

I cant figure out the V's of each node (I.E. Vp of EF86) because I dont know the bias of each tube.. I think that Ip EF86 should be at.. 2ma?

How do I choose my R's to lower the +B to 170V and 400V

thanks a lot!!!

 

[ray_moth]

How do I choose my R's to lower the +B to 170V and 400V

That is the least of your problems. Sorry, I realize that's not what you want to hear, but Claus Byrith's design as it stands is really not suitable for beam tetrodes like KT88. I don't think it's suitable even for EL34s! When Claus designed that modification to the Mullard 5-20, he got the fixed bias circuitry all wrong and he should have changed to a different tube for the splitter. Let me explain what I mean.

Firstly, regarding the bias circuit, think of what would happen if the wiper of the bias balancing adjustment pot were to go open circuit (pots are not perfect). The negative bias voltage would entirely disappear from the OP tube grids but there would still be a path to ground via the two 47k resistors. The OP tubes would quickly destroy themselves! In addition, the bias voltage level adjustment pot needs a padding resistor between the end of its track and ground, to limit its operation to a safe range. A more sensible biasing circuit is called for.

Secondly, the ECC83/12AX7 is not really suitable for driving EL34s with fixed bias, because of the requirement for reduced OP tube grid resistors. I don't think Claus lowered them enough, even for the EL34. However, in the case of KT88 (or any other beam tetrode) using fixed bias, those resistors must be no more than 100k. This is well outside the capabilities of ECC83 to drive! You need a higher current, lower internal resistance tube like 6CG7 or 6FQ7. (Better not to use ECC82 because, although it could do the job, it will degrade the sound).

Thirdly, a higher current, lower internal resistance splitter will also have lower amplification factor. 6CG7/6FQ7 will have only 20% of the gain of the ECC83 and it's probable that the loop gain of the amp will then be too low for effective NFB with adequate input sensitivity. However, you could remedy that satisfactorily if you revert to pentode operation of the EF86 first stage, which is what I would do.

 

[Eli Duttman]

Let me augment Ray's remarks. The tube used in a differential splitter should have a high mu for good phase to phase balance. The requirements (high gm and low Rp) that Ray outlined still hold. The 12AT7/ECC81 is the necessary combination of high mu, high gm, and low Rp. 200-230 V. on the 'T7 plates and Ib = 3 mA. sounds good.

I agree with Ray about using the EF86 in pentode mode, as it's very linear. Gas discharge regulator tubes are a way to get the B+ down for feeding the small signal pentodes. In addition, PSRR becomes a non-issue.


Edit: fixed typos

 

[ray_moth]

12AT7/ECC81 is the necessary combination of high mu, high gm, and low Rp

Eli's right, I forgot about that one.

Gas discharge regulator tubes are a way to get the B+ down for feeding the small signal pentodes. In addition, PSRR becomes a non-issue.

Excellent idea!

 

[zafira1981]

thank you very much both of you. I will modify the schematics and will post it so you let me know what you think.

I dont understand what happens if I change the 390K for 100K R in the bias circuit of the KT88's.


And about the fixed-bias pot for the KT..............................



when I first connected the +B in the output transformer, the KT's became RED and 1.5A passed through both KT's.....
only for a few seconds but....... very stupid from me and from the one who designed the circuit

thanksssssssss

 

[zafira1981]

Here is my new schematic.


Ra,Rb... Rk; C1-C6,CF (feedback) are values that I think I should modify and some that I dont know how to choose.

An externally hosted image should be here but it was not working when we last tested it.

thankssss

 

[Eli Duttman]

High gm types, like the 12AT7, are vulnerable to parasitic oscillation. Put a 1 KOhm Carbon composition "stopper" on each of the 'T7 grids.

Ray knows the "classic" Mullard circuit better than I do. Wait for his critique.

 

[ray_moth]

Hi, zafira1981.

I can make a few suggestions that may improve what you have there.

1. Do as Eli suggests and use stoppers on the grids of the splitter, replacing the 1k resistor from the plate of the EF86.

2. RA and RB seem to serve no purpose. Are they meant to drop the voltage from B+? If so, they are on the wrong side of the decoupling caps.

3. The splitter will work better if Rd is replaced by a CCS. A two-transistor cascode (like in Gingertube's 'Baby Huey' design) or a ring of two will do the job. Set it to supply 6mA (3mA each half, as Eli suggests). With a CCS in the tail, you can do away with the 50k pot for balancing the splitter. AC balance will then depend entirely on matching of the plate loads.

4. The voltage at the plate of the EF86 should be about 90v. The cathodes of the 12AT7 will be slightly higher (say 95v). To get 200v plate-cathode on the 12AT7 and reasonable voltage swing, the B+ for the splitter could be 450v. With 3mA per plate (set by the CCS), the plate load resistors Rc should then be 56k each. As mentioned above, these need to be well matched, maybe 2% tolerance.

5. Value of Re is a bit high at 100k, because it will be acting in series with Rg, approx. 0.5 x Rj, approx 0.5 x Rk and Rl. (Remove Rh, they shouldn't even be there!) I would reduce the value of each Re to 56k and make them 2% tolerance too, because they act in parallel with the splitter plate loads and will affect its AC balance. If you can reduce the bias balance adjustment pot Rj, so much the better. (This type of bias circuit is not my personal favorite; I prefer a separate pot to adjust each tube.)

6. As a safety measure in case of pot failure, you could connect a 220k resistor from each end of Rj to the -80v supply. That will guarantee a negative grid voltage if either Rj or Rk wiper goes open circuit. Of course, it won't work under those conditions because it will be cut off, but at least it will be safe.

 

[Jaap]

Ray

is this the bias scheme you propose ?

Why should Rh be left out ? Everybody has it in

 

[zafira1981]

Hi, zafira1981.

I can make a few suggestions that may improve what you have there.

1. Do as Eli suggests and use stoppers on the grids of the splitter, replacing the 1k resistor from the plate of the EF86.

2. RA and RB seem to serve no purpose. Are they meant to drop the voltage from B+? If so, they are on the wrong side of the decoupling caps.

3. The splitter will work better if Rd is replaced by a CCS. A two-transistor cascode (like in Gingertube's 'Baby Huey' design) or a ring of two will do the job. Set it to supply 6mA (3mA each half, as Eli suggests). With a CCS in the tail, you can do away with the 50k pot for balancing the splitter. AC balance will then depend entirely on matching of the plate loads.

4. The voltage at the plate of the EF86 should be about 90v. The cathodes of the 12AT7 will be slightly higher (say 95v). To get 200v plate-cathode on the 12AT7 and reasonable voltage swing, the B+ for the splitter could be 450v. With 3mA per plate (set by the CCS), the plate load resistors Rc should then be 56k each. As mentioned above, these need to be well matched, maybe 2% tolerance.

5. Value of Re is a bit high at 100k, because it will be acting in series with Rg, approx. 0.5 x Rj, approx 0.5 x Rk and Rl. (Remove Rh, they shouldn't even be there!) I would reduce the value of each Re to 56k and make them 2% tolerance too, because they act in parallel with the splitter plate loads and will affect its AC balance. If you can reduce the bias balance adjustment pot Rj, so much the better. (This type of bias circuit is not my personal favorite; I prefer a separate pot to adjust each tube.)

6. As a safety measure in case of pot failure, you could connect a 220k resistor from each end of Rj to the -80v supply. That will guarantee a negative grid voltage if either Rj or Rk wiper goes open circuit. Of course, it won't work under those conditions because it will be cut off, but at least it will be safe.

Thanks a lot. I'm drawing the schematic in VISIO. When I have it done, I will post it. thanks!!

 

[zafira1981]

1. Do as Eli suggests and use stoppers on the grids of the splitter, replacing the 1k resistor from the plate of the EF86.

Added 2x1K grid stoppers on each grid of 12AT7

2. RA and RB seem to serve no purpose. Are they meant to drop the voltage from B+? If so, they are on the wrong side of the decoupling caps.

Yes, they're meant to drop the voltage and the caps where wrong placed. If its not neccesary, .. no RA and RB

3. The splitter will work better if Rd is replaced by a CCS. A two-transistor cascode (like in Gingertube's 'Baby Huey' design) or a ring of two will do the job. Set it to supply 6mA (3mA each half, as Eli suggests). With a CCS in the tail, you can do away with the 50k pot for balancing the splitter. AC balance will then depend entirely on matching of the plate loads.

Good!!! I wanted to use a current-source but lot of people told me not to do that... well, I dont matter, I will try. For the - source I can rectify the 6.3v only for this, what do you think?

4. The voltage at the plate of the EF86 should be about 90v. The cathodes of the 12AT7 will be slightly higher (say 95v). To get 200v plate-cathode on the 12AT7 and reasonable voltage swing, the B+ for the splitter could be 450v. With 3mA per plate (set by the CCS), the plate load resistors Rc should then be 56k each. As mentioned above, these need to be well matched, maybe 2% tolerance.

OK!!

5. Value of Re is a bit high at 100k, because it will be acting in series with Rg, approx. 0.5 x Rj, approx 0.5 x Rk and Rl. (Remove Rh, they shouldn't even be there!) I would reduce the value of each Re to 56k and make them 2% tolerance too, because they act in parallel with the splitter plate loads and will affect its AC balance. If you can reduce the bias balance adjustment pot Rj, so much the better. (This type of bias circuit is not my personal favorite; I prefer a separate pot to adjust each tube.)

I changed the bias circuit.

6. As a safety measure in case of pot failure, you could connect a 220k resistor from each end of Rj to the -80v supply. That will guarantee a negative grid voltage if either Rj or Rk wiper goes open circuit. Of course, it won't work under those conditions because it will be cut off, but at least it will be safe.

OK

Here is the new schematic

An externally hosted image should be here but it was not working when we last tested it.

Thanks for your help!!!!

 

[ray_moth]

For the - source I can rectify the 6.3v only for this, what do you think?

There's no need for a negative supply for the CCS. The splitter cathodes are already at ~+95v which is plenty. The CCS can simply be connected between the cathodes and ground.

The rest of it looks OK at first glance but I don't have much time to think abour it right now. I'll get back to it later. I'd like to try modeling it with LTSpice to see if it's OK (the coward's way out!)

 

[zafira1981]

I dont have the LTSpice............

its *cof*COF* available for down*cof*cof*load via emule*cofcof* ?

 

[crispycircuit]

I have Dyna MK 3 power and output trannys waiting for a project just like this! I will join you in building a pr. of mono amps. I'm also real curious to see your power supply. I would like to use some heavy duty diodes for B+ and also run in pentode mode.... Any thoughts? I have a pair of AR11s (like AR3a) from the 70s and could use some power with some damping....??? Great Thread!!

 

[zafira1981]

OK, my power supply is:

8x220uF 400V for the B+
2x 0.47uF 630V
4xBY228 fast diodes (1.5KV / 2A)

PSU choke: 10H 250mA

PSU main transformer:
200+200AC,
70AC bias,
6.3V AC


The irons:

An externally hosted image should be here but it was not working when we last tested it.

The psu board

An externally hosted image should be here but it was not working when we last tested it.

An externally hosted image should be here but it was not working when we last tested it.

The layout

An externally hosted image should be here but it was not working when we last tested it.

An externally hosted image should be here but it was not working when we last tested it.

I got the 12AT7 (electro-harmonix Fender) so I'm building the new schematic rigth now.

 

[crispycircuit]

Thanks for the pics. Nice trannys! It's so cool to actually see how the assembled amp is done. Any thoughts of pentode drive? ..... And gas tube regulator?.... Draw up your power supply and lets see what the guru's say?? This forum has the most informed members I've seen. I got to buy some more books to get in with these tube guru's......

 

[frank754]

The output transformers should be at a 90 degree angle from the power xfmr, some people also use 45 degrees, but not in a line as you show.

 

[zafira1981]

Thanks for the pics. Nice trannys! It's so cool to actually see how the assembled amp is done. Any thoughts of pentode drive? ..... And gas tube regulator?.... Draw up your power supply and lets see what the guru's say?? This forum has the most informed members I've seen. I got to buy some more books to get in with these tube guru's......

Thanks!!!! I dont know about gas tube regulators. I will look for some info XD i have looooooot of old tubes in my depot, sure lot of gas tube regulators.

Tomorrow I will draw the psu.

The output transformers should be at a 90 degree angle from the power xfmr, some people also use 45 degrees, but not in a line as you show.

Yes, I know, but when I designed the layout I didnt know. What kind of trouble could that make? thanks

 

[refference]

Yes, I know, but when I designed the layout I didnt know. What kind of trouble could that make? thanks

Hi Zafira ,

The power transformer may induce "hummmmm" on the output
transformers , if all of them are "paralleled" .

If you put them in a right angle ( 90º ) , the magnetic field
of each one will not interact with another's magnetic field .

Regards,

Carlos

 

[coolmaster]

The output transformers should be at a 90 degree angle from the power xfmr, some people also use 45 degrees, but not in a line as you show.

May I know what is the reason or theory about this?
I'm thinking it applies only for a stereo block, should the mains tranny sit between 2 output trannnys.
I reckon its fine when HJ Leak monoblocks (or some others) have the 2 trannys facing the same direction but placed far away on each corner or side.