I designed an Amp using four pairs of push pull EL95 tubes, four in Parallel on each side of the push pull configuration.
Raa for a Pair of EL95 tubes is approx 10K so I opted for an Raa transformer 2.5K since I would have four pairs in parallel (with 43% UL taps) - ok so far.
B+ was set to exactly 300volts using a 6AS7G tube regulator circuit.
The amplifier works perfectly, good frequency response flat about 20KHz and below 15Hz. The problem is the power output. What ever I do, I cannot get more than 7W rms out of it before simple symmetrical clipping (20V pk - pk across my 8ohm test load). This is one quarter of what I expected.
Sowter transformers made the output transformers for me, they are 40W at 18Hz capable, so a lot of iron here. Brian Sowter and I discussed the Raa requirements in detail prior to the order.
It can't be a coincidence the power is 1/4 of what I expect with four pairs, either my design assumption is wrong about Raa or the transformer is wrong. All tubes are working, since unplugging pairs, drops the power evenly. I am using simple cathode bias, approx 330R AC bypassed resistor on each tube. I have swapped the Cathodyne phase inverter with a long tailed pair, zero and approx 10db of feedback, all the same result 8W or 20v pk-pk.
Here is an EL95 datasheet, at 300v B+ and a 330R resistor I get around 11volts on the cathode, (I know this is slightly hot, but I have tried a 22mA current source, 470R resister, everything here, and it still clips at 20v pk-pk).
I am out of ideas... any advice would be appreciated...
I started with this and improved it:
[²Ä¹æ ¿Àµð¿À ÀÚÀÛµ¿È£È¸] ÀÚÀÛÀÎÀÇ »ç¶û¹æ [²Ä¹æ]¿¡ ¿À½ÉÀ» ȯ¿µÇÕ´Ï´Ù^^
Raa for a Pair of EL95 tubes is approx 10K so I opted for an Raa transformer 2.5K since I would have four pairs in parallel (with 43% UL taps) - ok so far.
B+ was set to exactly 300volts using a 6AS7G tube regulator circuit.
The amplifier works perfectly, good frequency response flat about 20KHz and below 15Hz. The problem is the power output. What ever I do, I cannot get more than 7W rms out of it before simple symmetrical clipping (20V pk - pk across my 8ohm test load). This is one quarter of what I expected.
Sowter transformers made the output transformers for me, they are 40W at 18Hz capable, so a lot of iron here. Brian Sowter and I discussed the Raa requirements in detail prior to the order.
It can't be a coincidence the power is 1/4 of what I expect with four pairs, either my design assumption is wrong about Raa or the transformer is wrong. All tubes are working, since unplugging pairs, drops the power evenly. I am using simple cathode bias, approx 330R AC bypassed resistor on each tube. I have swapped the Cathodyne phase inverter with a long tailed pair, zero and approx 10db of feedback, all the same result 8W or 20v pk-pk.
Here is an EL95 datasheet, at 300v B+ and a 330R resistor I get around 11volts on the cathode, (I know this is slightly hot, but I have tried a 22mA current source, 470R resister, everything here, and it still clips at 20v pk-pk).
I am out of ideas... any advice would be appreciated...
I started with this and improved it:
[²Ä¹æ ¿Àµð¿À ÀÚÀÛµ¿È£È¸] ÀÚÀÛÀÎÀÇ »ç¶û¹æ [²Ä¹æ]¿¡ ¿À½ÉÀ» ȯ¿µÇÕ´Ï´Ù^^
Last edited:
20volts peak-peak = approx 8volts RMS = 8 watts
just corrected the typo in my original post!
I have some 1:10 probes and my scope is good for 400volts pk... or i could put my signal generator on the spare transformer i have and measure it that way I guess.... but i think I am missing something here...
I replaced the phase splitter with a long tailed pair from a donor leak Stereo twenty hot wired in. The phase splitter is having a hard time driving 4 pairs of tubes, and maybe I should up the 1K grid stoppers to something a lot more sensible. The concertina splitter is running off a seperate 250v supply... I'll reassemble the concertina splitter and put it back in circuit, but having got the same problem with a completely different known good splitter I am not convinced. As the input is turned up with the output clipping, the phase splitter drive carries on increasing, right up to the point where you even get crossover distortion appearing on the output despite the fact it is clipped. It seems like the OPT is incapable of giving more swing ...
The phase splitter signal does distort, once the output clips, but it still has more drive despite the distortion, leaving that issue aside it still looks like a problem with the transformer, but I am at a loss to prove this...
is 2.5K Raa a good choice with 4 paralleled 10k Raa tube pair? Brian Sowter at sowter transformers (world leaders IMHO) agreed it was but all these test results appear to say otherwise.
I would agree. The same Uaa as with one pair, but four times the current. So a quarter of 10k seems fine.
But ist your OT really what it should be., a 2.5k to 8R?
Do you get the swing on the anodes that you expect?
I recently had a similar problem with an EL34 PP. A pair of Electro Harmonics would not put ot the expected power. I tried a pair of JJ E34L and they were better.
The EHs were saturating, Ub was on the low side an Rc a bit high.
But ist your OT really what it should be., a 2.5k to 8R?
Do you get the swing on the anodes that you expect?
I recently had a similar problem with an EL34 PP. A pair of Electro Harmonics would not put ot the expected power. I tried a pair of JJ E34L and they were better.
The EHs were saturating, Ub was on the low side an Rc a bit high.
not that I know much about it..... but I just read that the EL95 should put out less power than EL84, and still even distorts more
About half the power (7 watts) for a PP class AB pair according to the data sheet.
jeff
not that I know much about it..... but I just read that the EL95 should put out less power than EL84, and still even distorts more
how about triode coupled ?
Its all about how you use them, I have approx 25 Mullard EL95 NOS sat in my valve-bits box and it seemed silly not to use them...
The amp works well up to 8Watts, really clean, but all that power and glass and iron in the OPT just for 8W something is not right here...
I never made a complete amplifier. I had over 100,000 used tubes and spent about 5 years sorting through them to keep those useful for audio amps. There were about 1000 6AQ5's, some in very poor condition. The 6AQ5 (EL90) has been used in audio and TV service for years, so I picked out the ones that had all their pins in good condition and didn't look burnt or gassy.
As with other tubes I tested them in pairs in pentode mode, using seperate variable power supplies (plate, screen and bias) and a 6600 ohm OPT that feeds a selection of load resistors allowing a plate to plate load from about 1K to 10K and power levels up to 200 watts at 1KHz. No feedback was used. Drive comes from a board I made a few years ago. It was designed with mosfet follower outputs so it is capable of dirving ANYTHING. The design started here:
http://www.diyaudio.com/forums/tube...r-board.html?highlight=universal+driver+board
And perfected here:
http://www.diyaudio.com/forums/tube...ab2-amp.html?highlight=universal+driver+board
It is total overkill for these little tubes, but I have much bigger tubes to fry too. A pair of 6AQ5's will make 10 to 12 watts on 300 volts with an 8K ohm load. Slightly higher power can be had on higher voltage, but the 6AQ5 is easy to melt and there is only a small headroom between a slight red glow on the plate and a hole in the glass! I melted a dozen or so to figure this out.
Since I had several hundred tubes that looked rather ugly, I did wire a few together just to see what happens. It seems that the sweet spot is 6 or 8 tubes. I think most of my testing was done with 3300 ohms and maybe a bit over 300 volts. I have all of this written down somewhere, but I am a bit disorganized now.
Yes, I have done parallel sets in P-P and SE. You just divide the suggested load by the number of tubes (or pairs if P-P). I am working on a stereo amp using 16 13GB5's (8 per channel). A single pair worked well with 600 volts into a 5K ohm load. The monster (still unfinished) will use a 1.25K load on 600 to 650 volts to make 400 to 500 WPC.
Just brain storming your problem, reproduced here for comment:
2500:8 => Impedance ratio of 312.5
Turns ratio is therefore 17.7
20V pk to pk at secondary means 20 x 17.7 = 354V pk-pk across full primary
That is 177V pk across each side of the push pull.
With 300 rail you should be able to get:
300 -11 V at cathode gives 289 across the tube.
EL95 saturation voltage will be around 40V
So should be able to swing 289 - 40 = 249 volts at the anode
or 2 x 249 = 498V pk-pk across the full primary
Which would give 498/17.7 = 28.1 V pk-pk across the 8 Ohm secondary or 9.9V RMS or 12.3 Watts into 8 Ohms maximum.
I think that 300V rail is too low but even with 300V you should be getting 28 volts pk-pk at the secondary before clipping.
Is your regulated supply dropping at higher current peaks?
Cheers,
Ian
2500:8 => Impedance ratio of 312.5
Turns ratio is therefore 17.7
20V pk to pk at secondary means 20 x 17.7 = 354V pk-pk across full primary
That is 177V pk across each side of the push pull.
With 300 rail you should be able to get:
300 -11 V at cathode gives 289 across the tube.
EL95 saturation voltage will be around 40V
So should be able to swing 289 - 40 = 249 volts at the anode
or 2 x 249 = 498V pk-pk across the full primary
Which would give 498/17.7 = 28.1 V pk-pk across the 8 Ohm secondary or 9.9V RMS or 12.3 Watts into 8 Ohms maximum.
I think that 300V rail is too low but even with 300V you should be getting 28 volts pk-pk at the secondary before clipping.
Is your regulated supply dropping at higher current peaks?
Cheers,
Ian
The regulated supply is rock steady, doesn't drop a single volt. But you have answered the question. I will never get more than about 12W from this design, so I either up the HT to around 400 (not sure an EL95 will like that) or have a 4-5K transformer which will be out of tune, but give me more swing and a few more watts...
Given these calculations it seems almost pointless adding tubes in Parallel unless you can up the HT...
Given these calculations it seems almost pointless adding tubes in Parallel unless you can up the HT...
You should verify that the turns ratio of the transformer is really as it should be.
Could you remove it from the circuitry and by using an audio generator feed 1 kHz AC signal to anode taps and then measure the voltage at the 8 ohms output taps. The ratio of these two voltages represents the turns ratio and should be around 18.
EL95 datas are given as pentode connection, not UL. Try if you get essential difference when UL is changed to pentode.
Could you remove it from the circuitry and by using an audio generator feed 1 kHz AC signal to anode taps and then measure the voltage at the 8 ohms output taps. The ratio of these two voltages represents the turns ratio and should be around 18.
EL95 datas are given as pentode connection, not UL. Try if you get essential difference when UL is changed to pentode.
Last edited:
Gingertube, something in your brain storming confused me.
To confirm what you did I re-engineered the single pair pp- circuit of the Philips data sheet. It runs of Ub=250V and puts out 7W.
10k : 8R - turns ratio 35.4
voltage swing 250V - 10V - 40V = 200V
double that: 400V
divide by turns ratio 11.3Vpp => 4Vrms => 2W @ 8R
But Philips say you get 7W in that setup.
I think the voltage over one tube is 200Vp, not 200Vpp.
Assuming that, I get 400Vpp over one tube, the voltage on the other tube is 400Vpp as well, but 180 deg. out of phase. So in total I get 800Vpp a-a.
800V divided by turns ratio gives: 22.6Vpp => 8Vrms => 8W @ 8R
Maybe the tubes won´t saturate as nicely or there may be some other losses, so the claim of 7W seems realistic.
And the OP´s output stage?
(300V - 40V -10V)*4 = 1000Vpp
1000V/17.7 = 56.5Vpp => 20Vrms => 50W @8R
In real life he should be able to get 40W. That is more than 4 times what the data sheet says for one pair, but that is due to his using a higher Ub.
Peterfarrow, have you checked the turns ratio of yout OT yet?
To confirm what you did I re-engineered the single pair pp- circuit of the Philips data sheet. It runs of Ub=250V and puts out 7W.
10k : 8R - turns ratio 35.4
voltage swing 250V - 10V - 40V = 200V
double that: 400V
divide by turns ratio 11.3Vpp => 4Vrms => 2W @ 8R
But Philips say you get 7W in that setup.
I think the voltage over one tube is 200Vp, not 200Vpp.
Assuming that, I get 400Vpp over one tube, the voltage on the other tube is 400Vpp as well, but 180 deg. out of phase. So in total I get 800Vpp a-a.
800V divided by turns ratio gives: 22.6Vpp => 8Vrms => 8W @ 8R
Maybe the tubes won´t saturate as nicely or there may be some other losses, so the claim of 7W seems realistic.
And the OP´s output stage?
(300V - 40V -10V)*4 = 1000Vpp
1000V/17.7 = 56.5Vpp => 20Vrms => 50W @8R
In real life he should be able to get 40W. That is more than 4 times what the data sheet says for one pair, but that is due to his using a higher Ub.
Peterfarrow, have you checked the turns ratio of yout OT yet?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.