I disagree about fixed bias. Cathode bias, and individual cathode resistors and decoupling capacitors will give the best results. The last thing you want to do with PPP is be fiddling with lots of bias points for lots of tubes. If you go ppp, half the Raa for two pairs of tubes, one third Raa for three pairs and so on. There is a limiting factor though in playing this game, and that is you will eventually run into a limit on power output imposed by the B+. You might find in theory you should get the power, as the load impedance is lower but in practice in just doesn't work well, until you start pushing B+. I've designed and built lots of PPP amplifiers, my latest one uses eight 6v6 tubes per channel in UL mode. You will find UL mode is half way between Triode and Pentode, and generally gives the lowest distortion. If your output transformer is high quality the voltage drop from the screen taps to the anode taps from the B+ will be negligible. This means you should look at your curves and assume vg2=B+= anode voltage. #
If you choose your transformer so it has half the impedance Raa than for a single pair (for two pairs) then you won't go wrong using this for UL as well. In theory the "perfect" Raa will be less in UL mode, but it also depends on the Bias points for the tubes. If the tubes are bias for AB operation (good practice for efficiency and power) then during the class A operation the Raa is what you expect, but when the amp goes into class B at higher power, one tube cuts off, and the other only sees half the transformer as a load, half the transformer = half the turns, half the turns = 1/4 the impedance.
Don't get hung up on this, from the 6L6 datasheet, choosing self (Cathode bias) and a B+ pf 400v and a screen of 300volts Raa= 6600 and PO=32watts @ 2%. In ultra linear mode, you will have to drop the B+ to 300v, the screen will also be 300volts (approx) and your power output will be around 20watts, two pairs will give you around 40Watts, and I would choose an Raa of around 3K. Since the screen rating on the 6L6 is low, you are better off choosing KT66 tubes, GEC specified a load of 40% UL operation at 425 va.Vg2 giving 32Watts RMS and an Raa of 7K, two pairs with a load of 3.5K at this B+ will give you over 60Watts, thats 50% more than the 6L6. My 6v6 x eight PPP 43% taps gives 61watts RMS and distortion figure of under 0.3% measured on a keithley 2015.
Cathode bias is strongly recommended though, unless you want to spend your life adjusting trim pots and measuring current. I also use an ECC82 cathode follower between the phase splitter and the output tubes. Miller effect is not a practical issue with pentodes , even in UL down to 40% taps. You will need a cathode follower though.
If you choose your transformer so it has half the impedance Raa than for a single pair (for two pairs) then you won't go wrong using this for UL as well. In theory the "perfect" Raa will be less in UL mode, but it also depends on the Bias points for the tubes. If the tubes are bias for AB operation (good practice for efficiency and power) then during the class A operation the Raa is what you expect, but when the amp goes into class B at higher power, one tube cuts off, and the other only sees half the transformer as a load, half the transformer = half the turns, half the turns = 1/4 the impedance.
Don't get hung up on this, from the 6L6 datasheet, choosing self (Cathode bias) and a B+ pf 400v and a screen of 300volts Raa= 6600 and PO=32watts @ 2%. In ultra linear mode, you will have to drop the B+ to 300v, the screen will also be 300volts (approx) and your power output will be around 20watts, two pairs will give you around 40Watts, and I would choose an Raa of around 3K. Since the screen rating on the 6L6 is low, you are better off choosing KT66 tubes, GEC specified a load of 40% UL operation at 425 va.Vg2 giving 32Watts RMS and an Raa of 7K, two pairs with a load of 3.5K at this B+ will give you over 60Watts, thats 50% more than the 6L6. My 6v6 x eight PPP 43% taps gives 61watts RMS and distortion figure of under 0.3% measured on a keithley 2015.
Cathode bias is strongly recommended though, unless you want to spend your life adjusting trim pots and measuring current. I also use an ECC82 cathode follower between the phase splitter and the output tubes. Miller effect is not a practical issue with pentodes , even in UL down to 40% taps. You will need a cathode follower though.
I am for the DC coupled follower. This can also be a vacuum tube in this case. It is so much better than any other drive/bias solution. There is no need for individual bias. The devices can be biased in pairs.
Having 8 devices, if one asks for a good matching ( 8 devices where there is a difference within 10% between the extremes, for example, and even better if 5%) it is relatively easy to find pairs for which PPP distortion is minimized and way better than the PP at any practical power level.
The Hammond 1650 with 1.9K primary load I think is for the PPP with fixed bias in pentode connection where plate voltage is 360V and G2 at 270V.
Having 8 devices, if one asks for a good matching ( 8 devices where there is a difference within 10% between the extremes, for example, and even better if 5%) it is relatively easy to find pairs for which PPP distortion is minimized and way better than the PP at any practical power level.
The Hammond 1650 with 1.9K primary load I think is for the PPP with fixed bias in pentode connection where plate voltage is 360V and G2 at 270V.
I agree 45. A composite device by definition will give lower distortion. Matching in pairs is probably best but not essential.
Cathode bias in my opinion just wastes power and leads to an eventual failure when one tube starts drawing grid current. one idea already mentioned is a CCS in each cathode. This is also an excellent idea but is also wastefull. It forces class A which may or may not be desirable.
Overall no matter what the bias controls are I think controling grid current tightly is the answer and this means DC coupling. A side advantage is AB2 operation. In UL It really takes an optimal output transformer(super conducting?) not to drop the anode below the screen on the peaks. DC coupled triode with cathode feedback if you need it is the way to go IMHO.
Cheers
Matt.
Cathode bias in my opinion just wastes power and leads to an eventual failure when one tube starts drawing grid current. one idea already mentioned is a CCS in each cathode. This is also an excellent idea but is also wastefull. It forces class A which may or may not be desirable.
Overall no matter what the bias controls are I think controling grid current tightly is the answer and this means DC coupling. A side advantage is AB2 operation. In UL It really takes an optimal output transformer(super conducting?) not to drop the anode below the screen on the peaks. DC coupled triode with cathode feedback if you need it is the way to go IMHO.
Cheers
Matt.
You can use "iron" with a separate screen grid winding and regulate g2 B+, while still getting UL mode's local NFB.
Plitron's 100 W. rated PAT-2100-SSCR might be OK for the OP's purpose. Of course, ONetics, Heyboer, etc. can be commissioned to wind something custom.Agreed, UL is pretty good. Toroidal outputs with multiple output tubes would be a pain. A servo scheme would be nice and I have tried it. It does work very well but there is that niggle of what if it goes wrong..... Multiple chain reaction catastrophic failure. Custom EI or C core trannys with seperate UL windings really are the answer as you say. You guys in the US are lucky that way.
Cheers
Matt.
Cheers
Matt.
"Old Blighty" has its resources too. UK DIYers can contact Brian Sowter and find out what his family firm can do.
I'm building some parallel push pull KT88 with Plitron Torroidal Outut Trannies VDV-2100-CFB/H (10% catyhode feeback + 30% ultralinear).
To deal with the output tubes I did'nt muck about, I'm using one of Menno/Guido's bias servo boards in each monoblock
Tube bias control
@Kay Pirinha (pg 1) Parallel Push Pull (PPP) is usually used, these days, to denote an ordinary Push Pull Amp where the output tubes on each side have been doubled up. That is the tubes on each side of the Push Pull are treated as a "composite" single tube with double the gm and half the rp.
It is unfortunate that those who adopted this naming ignored the fact that there already was an entirely different circuit technique known as Parallel Push Pull where a single pair of push pull tubes are effectively run in AC parallel (Circlotron style circuit), So we just have to put up with "folk" stealing a circuit name and applyiing it to something quite different.
My introduction to the PPP circuits you are talking about was these:
http://www.elektor.com/magazines/2001/may/ppp-hifi-valve-power-amplifier.54477.lynkx
I'm not sure why this technique is'nt a lot more popular. As the article text says, it is a little less efficient than "ordinary" Class AB Push Pull, otherwise all the differences are positive.
Cheers,
Ian
To deal with the output tubes I did'nt muck about, I'm using one of Menno/Guido's bias servo boards in each monoblock
Tube bias control
@Kay Pirinha (pg 1) Parallel Push Pull (PPP) is usually used, these days, to denote an ordinary Push Pull Amp where the output tubes on each side have been doubled up. That is the tubes on each side of the Push Pull are treated as a "composite" single tube with double the gm and half the rp.
It is unfortunate that those who adopted this naming ignored the fact that there already was an entirely different circuit technique known as Parallel Push Pull where a single pair of push pull tubes are effectively run in AC parallel (Circlotron style circuit), So we just have to put up with "folk" stealing a circuit name and applyiing it to something quite different.
My introduction to the PPP circuits you are talking about was these:
http://www.elektor.com/magazines/2001/may/ppp-hifi-valve-power-amplifier.54477.lynkx
I'm not sure why this technique is'nt a lot more popular. As the article text says, it is a little less efficient than "ordinary" Class AB Push Pull, otherwise all the differences are positive.
Cheers,
Ian
Last edited:
Do you mind sharing the schematic for the PIC uC circuit? Sounds interesting.
My Radiotron designers handbook from 1951 describes parallel push pull as exactly what I am using here. I think that predates your description, so if there is any name "stealing" is the other way round.
P.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.