Phil,
You could try looking for a pair of second hand 8k transformers and wire them up using the 4 ohm tap. This arrangement is less than ideal but may well sound ok.
if you are not happy with the results you could ditch the EL42s and use the transformers with tubes suited to an 8k primary (eg EL84)
You could try looking for a pair of second hand 8k transformers and wire them up using the 4 ohm tap. This arrangement is less than ideal but may well sound ok.
if you are not happy with the results you could ditch the EL42s and use the transformers with tubes suited to an 8k primary (eg EL84)
One comment: I'm not sure you could get a 12AV7 to work properly in a direct-couple voltage amp/split-load because of the low available B+. If we run the plate voltage of the first stage as low as 100V (and that's getting down near where the tube will be pretty unhappy), the second stage will have about 102V on the cathode. Likewise, the drop across the plate resistor will be about 102V, so with a max of 250V available, you're stuck with under 50V across the tube. it will NOT be happy.
You've got four choices, as I see it:
1. Use a different topology. A long tail voltage amp/phase splitter could work on the available B+ with a 12AV7 or a 12AT7.
2. Couple with a step network, i.e., a voltage divider with the upper resistor bypassed with a capacitor. If you choose a divider ratio to get you 70V on the grid of the split load inverter, that leaves 110V across the tube in that section, a much more reasonable operating voltage.
3. Capacitor couple from the voltage amp to the split load. That gives the most flexibility, but you'll have to take care about low frequency stability since you've introduced a third LF rolloff.
4. Use a tube that can tolerate lower voltages. The ECC88 comes to mind. If you have a copy of Morgan Jones's "Valve Amplifiers," it's worth working through the Bevois Valley input stage calculations. If you don't have it, time to buy it.
You've got four choices, as I see it:
1. Use a different topology. A long tail voltage amp/phase splitter could work on the available B+ with a 12AV7 or a 12AT7.
2. Couple with a step network, i.e., a voltage divider with the upper resistor bypassed with a capacitor. If you choose a divider ratio to get you 70V on the grid of the split load inverter, that leaves 110V across the tube in that section, a much more reasonable operating voltage.
3. Capacitor couple from the voltage amp to the split load. That gives the most flexibility, but you'll have to take care about low frequency stability since you've introduced a third LF rolloff.
4. Use a tube that can tolerate lower voltages. The ECC88 comes to mind. If you have a copy of Morgan Jones's "Valve Amplifiers," it's worth working through the Bevois Valley input stage calculations. If you don't have it, time to buy it.
Thanks for the responses
Eli, hmmm that hammond transformer, the 'E' model, at least its not the smallest available... there are smaller and cheaper and although the frequency response is stated as 150 Hz - 15 Khz, this value is stated for the full 125 range, my thinking is that the larger, 15W model has to be better than the smallest 3W model, right?
I think regarding the OT situation that I will build the amplifier as planned leaving purchasing OT's until last, I've sent a few emails and I'm working through the transformer suppliers on the net, hopefully something of reasonable quality will come up second hand or just at a good price. In the mean time I have a working transformer that I can use to test the outputs one at a time which will be useful for de-bugging.
Mach1, regarding your comment about switching to EL84's in the future if I'm not happy with the OT's that I can get for the EL42's; could the rest of the circuit be left the same if this were the case? would my power transformers still be upto the job?
SY, thanks for the advice, option 2 seems the most appealing to me.... mainly because I understand it
do you mind me asking what you would do if you were me?
and yes I've seen many recommendations on this forum for the Morgan Jones book, it is time I got myself one!
Eli, hmmm that hammond transformer, the 'E' model, at least its not the smallest available... there are smaller and cheaper
and although the frequency response is stated as 150 Hz - 15 Khz, this value is stated for the full 125 range, my thinking is that the larger, 15W model has to be better than the smallest 3W model, right?I think regarding the OT situation that I will build the amplifier as planned leaving purchasing OT's until last, I've sent a few emails and I'm working through the transformer suppliers on the net, hopefully something of reasonable quality will come up second hand or just at a good price. In the mean time I have a working transformer that I can use to test the outputs one at a time which will be useful for de-bugging.
Mach1, regarding your comment about switching to EL84's in the future if I'm not happy with the OT's that I can get for the EL42's; could the rest of the circuit be left the same if this were the case? would my power transformers still be upto the job?
SY, thanks for the advice, option 2 seems the most appealing to me.... mainly because I understand it
do you mind me asking what you would do if you were me?and yes I've seen many recommendations on this forum for the Morgan Jones book, it is time I got myself one!
Unfortunately not - the EL84s will require more current and a bit more drive from the preceeding circuitry (from memory the EL42 biases very low).
I second the idea of purchasing 'Valve Amplifiers'. The step through Beauvois Valley amp design process will give you quite a good idea of the design principles involved.
SY is 100% correct about the 6DJ8/ECC88 working well at modest B+ rail voltages. 90 V. on the plate is "textbook". However the μ for the type is only 33. Could enough gain be obtained, without resorting to CCS loading?
I LIKE the 12AT7/ECC81 in PP amps. Pete Millett has provided us with the means to get a "tall" B+ supply for the 'T7, without adding any additional "iron". Some "sand" in the PSU gets the job done. Connect both plates of the EZ41 to the rectifier winding's CT. The B+ for the EL42s still gets taken from the EZ41's cathode. Connect a bridge rectifier made 4X UF4007s to the ends of the rectifier winding. Each UF4007 should be paralleled by a HIGH WVDC/10 nF. snubber capacitor. Ground the anode junction of the bridge. Take the "tall" B+ rail from the cathode junction of the bridge. The "short" B+ rail is full wave rectified, even though a quick look indicates otherwise. The path to ground for both rails is via the same 2X UF4007s. A Maida style regulator will set 12AT7/ECC81 B+ exactly to the value SY wants.
Things will be difficult enough using the 125E. In order to extend that 150 Hz. down to something reasonable, a good sized NFB loop error correction signal will be needed. Such an error correction signal requires magnetic headroom. The cores in the smaller 125 series trafos are too small. AAMOF, to further reduce the chance of O/P core saturation, change the 510 KOhm grid leak resistors to 470 KOhms. The high pass "corner" freq. moves up, but stays below 30 Hz. The little bit of "bending" being done to the 1:10 impedance rule is probably of no consequence.
Speaking of NFB loop error correction signals, change the value of the cap. that connects the "hot" speaker wire to ground from 1200 pF. to 1500 pF. Miracles do not come forth from Hammond 125Es.
Converting to 100% SS B+ rectification might get the "short" B+ rail large enough to use EL821s as the O/P tubes and Edcor XPP15-8-8K O/P trafos. EL821s require 1 KOhm resistors in the triode tie positions.
I LIKE the 12AT7/ECC81 in PP amps. Pete Millett has provided us with the means to get a "tall" B+ supply for the 'T7, without adding any additional "iron".
Some "sand" in the PSU gets the job done. Connect both plates of the EZ41 to the rectifier winding's CT. The B+ for the EL42s still gets taken from the EZ41's cathode. Connect a bridge rectifier made 4X UF4007s to the ends of the rectifier winding. Each UF4007 should be paralleled by a HIGH WVDC/10 nF. snubber capacitor. Ground the anode junction of the bridge. Take the "tall" B+ rail from the cathode junction of the bridge. The "short" B+ rail is full wave rectified, even though a quick look indicates otherwise. The path to ground for both rails is via the same 2X UF4007s. A Maida style regulator will set 12AT7/ECC81 B+ exactly to the value SY wants.Things will be difficult enough using the 125E. In order to extend that 150 Hz. down to something reasonable, a good sized NFB loop error correction signal will be needed. Such an error correction signal requires magnetic headroom. The cores in the smaller 125 series trafos are too small. AAMOF, to further reduce the chance of O/P core saturation, change the 510 KOhm grid leak resistors to 470 KOhms. The high pass "corner" freq. moves up, but stays below 30 Hz. The little bit of "bending" being done to the 1:10 impedance rule is probably of no consequence.
Speaking of NFB loop error correction signals, change the value of the cap. that connects the "hot" speaker wire to ground from 1200 pF. to 1500 pF. Miracles do not come forth from Hammond 125Es.
Converting to 100% SS B+ rectification might get the "short" B+ rail large enough to use EL821s as the O/P tubes and Edcor XPP15-8-8K O/P trafos. EL821s require 1 KOhm resistors in the triode tie positions.
Use EDCOR model XPP15-4-8K O/P trafos in combination with 8 Ohm speakers. Hopefully, the primary inductance will be sufficient.Thanks Eli, I was actually thinking the same thing with the OT, the correct ratio between primary and secondary windings is the important thing and that would still be kept, I would be very happy to use the Edcor transformers
now as for all these other driver stage options its giving me a bit of a headache! A few thoughts tho... as I am planning to build both channels into the same chassis, if an extra gain stage is needed (say using ECC88s) then at least it would only be one extra tube and not two (half for each channel). in which case if there isn't enough spare power for the extra tube then there are two options as I see it, add a THIRD power transformer, which would only have to be a tiddler, or replace the old transformers with one big one thats got enough juice for everything... reducing the only usable parts from my amps to the EL42's and their sockets! but I'm ok with this.
If I do replace the power transformer this will also open up my options for driver tubes as I could use a higher suply than 250V.
personally I don't mind using diodes instead of tubes for rectifiers but I would prefer to keep the level of 'sand' to that, I'm sure the above circuit you mention could work but I would prefer to keep things simple
now as for all these other driver stage options its giving me a bit of a headache! A few thoughts tho... as I am planning to build both channels into the same chassis, if an extra gain stage is needed (say using ECC88s) then at least it would only be one extra tube and not two (half for each channel). in which case if there isn't enough spare power for the extra tube then there are two options as I see it, add a THIRD power transformer, which would only have to be a tiddler, or replace the old transformers with one big one thats got enough juice for everything... reducing the only usable parts from my amps to the EL42's and their sockets! but I'm ok with this.
If I do replace the power transformer this will also open up my options for driver tubes as I could use a higher suply than 250V.
personally I don't mind using diodes instead of tubes for rectifiers but I would prefer to keep the level of 'sand' to that, I'm sure the above circuit you mention could work but I would prefer to keep things simple
Re: Re: Design using EL42 output pentodes
Hiya Merlinb, nice to find a local on here I'm living in Caerphilly, thats about 10 miles outside of Cardiff. Thanks for the offer of practical help, very kind of you. I think to be honest once these design hurdles are over the practical part will hopefully be fairly straightforward 
Hiya Merlinb, nice to find a local on here
I'm living in Caerphilly, thats about 10 miles outside of Cardiff. Thanks for the offer of practical help, very kind of you. I think to be honest once these design hurdles are over the practical part will hopefully be fairly straightforward Phil, we'll keep things as simple as satisfactory performance permits.
Keep the OEM power trafo and B+ rectifier topology. You will use a 6DJ8/ECC88 family tube in the small signal "hole". The current production JJ E88CC/6922 is both decent and affordable.
Remember my previous remark about sufficient gain. CCS, instead of resistive (R7), loading will be used on the voltage amplifier. The least complicated CCS is the IXYS 10M45S. Let's look at the E88CC data sheet. If IB is forced to be 4 mA. by the CCS and -3 V. are on the grid, 110 V. are on the plate. That's a very suitable operating point. Ohm's Law tells us that 750 Ohms drop 3 V., when the current present is 4 mA. 100 of those Ohms are already accounted for by R6. Set R5 as close to 650 Ohms as you can, in an available part. C2 should be 100 μF. and not less than 6 WVDC. Now we wait for SY or other knowledgeable person to compute the R9/R11 value.
R9/11 are a piece of cake to compute. Select a current for the second stage- I'd probably use 6 or 7 mA, it's not that critical. The cathode will be just a couple volts north of the grid. With 110V on the plate of the voltage amp, the cathode voltage of the split load will be roughly 112V. That also means that the drop across the plate resistor is 112V, so you've only got 250 - (2x112) = 26V across the tube. Tilt!
Let's try it again. Redo the current in the first stage so that its plate voltage is closer to 80V. That means 82V on the split load cathode, 82V drop across the plate resistor. Now we have 250 - (2x82) = 86V across the split load tube. MUCH better.
The split load resistors would then be R = V/I = 82V drop/6mA current = 13.7k. Anything from 12-15k will work fine, but the two resistors should be closely matched.
If you want to be a little tricky in the first stage, substitute a red LED (1.7V drop) for R5/C2.
Let's try it again. Redo the current in the first stage so that its plate voltage is closer to 80V. That means 82V on the split load cathode, 82V drop across the plate resistor. Now we have 250 - (2x82) = 86V across the split load tube. MUCH better.
The split load resistors would then be R = V/I = 82V drop/6mA current = 13.7k. Anything from 12-15k will work fine, but the two resistors should be closely matched.
If you want to be a little tricky in the first stage, substitute a red LED (1.7V drop) for R5/C2.
My gut feeling is that "less is more", in this case. How much current can safely be pulled from the B+ PSU? Remember, the OEM small signal tube is a 12AX7/ECC83. Would 18 KOhm "concertina" resistors be acceptable? Reworking the filter the rectifier drives into a CLC with 10 μF. in the 1st position followed by a low DCR 5 H. inductor should keep the rail voltage up, while allowing a little more current to be squeezed out. Maximize small signal B+ voltage by using a "check valve" style decoupling network of 22 Ohms and UF4007 in series followed by a 15 μF. capacitor.
Does IB = 3 mA. in combination with 2.5 V. of bias meet with your approval? 2.5 V. is already too low for handling the full 2 VRMS O/P of a "standard" CDP. An 18 KOhm resistor between the I/P and the "top" of the volume control deals with that. However, if the padding resistor becomes larger, we might run out I/P drive.
What's the best compromise?
Phil,
SY has finished things off very nicely for you. Per his remark about I/P voltage swing limit, forget a padding resistor at the "top" of the volume control.
IB = 3 mA. and bias at -2.5 V. make R5 = 733 Ohms. C2 remains 100 μF.
Use 1% 18.2 KOhm 1/2 W. metal film parts for R9 and R11.
A Hammond 156L or something similar is appropriate for the B+ PSU filter choke.
It seems you get to recycle all 8 tube sockets and both power trafos.
Please draw up a revised schematic and post it.
SY has finished things off very nicely for you. Per his remark about I/P voltage swing limit, forget a padding resistor at the "top" of the volume control.
IB = 3 mA. and bias at -2.5 V. make R5 = 733 Ohms. C2 remains 100 μF.
Use 1% 18.2 KOhm 1/2 W. metal film parts for R9 and R11.
A Hammond 156L or something similar is appropriate for the B+ PSU filter choke.
It seems you get to recycle all 8 tube sockets and both power trafos.
Please draw up a revised schematic and post it.
Ok here is the revised schematic as I understand it, I think there is a good chance I've misinterpreted you here, with the power supply and CCS... sorry if I have. I've also assumed those changes you mentioned to the feedback loop if using the hammond OPT aren't necessary now, correct me if I'm wrong.
Phil,
R2 = 470 KOhms and C8 = 1800 pF. Those values will give a little added protection against O/P trafo core saturation and small signal circuitry slew limiting. 470 KOhms combines with 12 nF. to give a 3 dB. down point of 28.2 Hz. IIRC, that's below the lowest note a double bass can produce. 1100 Ohms combines with 1800 pF. to "corner" at 80.3 KHz. Stray wiring capacitance will lower that value slightly. Still, for "Red Book" CD as the signal source, things are fine. CCS loading the voltage gain triode prevents the use of "peaking". Let's avoid a problem in trying to do too much. "Perfect" phase compensation requires the use of an oscilloscope. This circuit gets pretty good via brute force. Thankfully, the 6922/E88CC has a high gm.
The schematic symbol for a CCS is a pair of interlocked circles. You state the current present, which in this case is 3 mA. You need a few passive parts in combination with a 10M45S to set the current. Please refer to the data sheet I previously linked.
The B+ PSU filter is CLC. A 47 μF. cap. follows the 5 H. choke.
The rectifier's heater wiring is yet to be corrected. None of the heaters get connected directly to the chassis.
BTW, use a good quality, low impedance, 'lytic for C2. Panasonic and Nichicon offer suitable parts. A low quality part will not properly transmit the high freq. error correction signal.
R2 = 470 KOhms and C8 = 1800 pF. Those values will give a little added protection against O/P trafo core saturation and small signal circuitry slew limiting. 470 KOhms combines with 12 nF. to give a 3 dB. down point of 28.2 Hz. IIRC, that's below the lowest note a double bass can produce. 1100 Ohms combines with 1800 pF. to "corner" at 80.3 KHz. Stray wiring capacitance will lower that value slightly. Still, for "Red Book" CD as the signal source, things are fine. CCS loading the voltage gain triode prevents the use of "peaking". Let's avoid a problem in trying to do too much. "Perfect" phase compensation requires the use of an oscilloscope. This circuit gets pretty good via brute force. Thankfully, the 6922/E88CC has a high gm.
The schematic symbol for a CCS is a pair of interlocked circles. You state the current present, which in this case is 3 mA. You need a few passive parts in combination with a 10M45S to set the current. Please refer to the data sheet I previously linked.
The B+ PSU filter is CLC. A 47 μF. cap. follows the 5 H. choke.
The rectifier's heater wiring is yet to be corrected. None of the heaters get connected directly to the chassis.
BTW, use a good quality, low impedance, 'lytic for C2. Panasonic and Nichicon offer suitable parts. A low quality part will not properly transmit the high freq. error correction signal.
Thanks Eli
I mistakenly took R5 and R6 as providing the resistance referred to as Rk on the data sheet. I have added Rk, which from the graph = 1140 Ohms. The data sheet does not show any other parts, however you said "You need a few passive parts in combination with a 10M45S to set the current" Please let me know if anything else is needed. I have kept the incorrect symbol for the 10M45S for now just for clarity for myself when I'm connecting it up.
What would you suggest for a good cap. type for C1?
Also you mentioned the voltage rating for C2 to be not less than 6V, am I right in thinking this was a typo? I've not seen such a low voltage rating on an electrolytic.
I mistakenly took R5 and R6 as providing the resistance referred to as Rk on the data sheet. I have added Rk, which from the graph = 1140 Ohms. The data sheet does not show any other parts, however you said "You need a few passive parts in combination with a 10M45S to set the current" Please let me know if anything else is needed. I have kept the incorrect symbol for the 10M45S for now just for clarity for myself when I'm connecting it up.
What would you suggest for a good cap. type for C1?
Also you mentioned the voltage rating for C2 to be not less than 6V, am I right in thinking this was a typo? I've not seen such a low voltage rating on an electrolytic.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.