No matching is necessary, but I have found minor improvements in distortion performance (usually not enough to be noticeable) from swapping tubes around. 6GU5's are pretty common, but 6HJ5's can be scarce. I would get at least one spare set. If you do this, you can match tubes in the amp by picking pairs that have the lowest distortion. If no measuring equipment is available, pick tubes that reach the desired current (30 to 35 mA) at similar grid voltages.
Last time I checked vacuumtubes.net had considerable stock of both tubes.
Last time I checked vacuumtubes.net had considerable stock of both tubes.
Thank you both on clarification of matching.
George, I'd like to replicate your 125W build, especially the one you mentioned with a maggie, after read your previous post l understood that you had 300V(DC or AC?) on driver board, but another separate 600V(again DC or AC?) on OP transformer B+ connector, is it correct? Thanks
George, I'd like to replicate your 125W build, especially the one you mentioned with a maggie, after read your previous post l understood that you had 300V(DC or AC?) on driver board, but another separate 600V(again DC or AC?) on OP transformer B+ connector, is it correct? Thanks
We use two 300 volt DC supplies stacked on top of each other (in series).
The lower supply IS the power supply built into the board with upgraded diodes to handle the extra current. With a power transformer in the 230 to 250 volt range the raw B+ will be around 300 volts. This enough to feed the on board screen regulator at 150 volts, and the filtered B+ which will be 5 to 10% lower than the raw B+. Everything except for the plates of the output tubes amp runs off of this supply as it did in Pete's version, but at a slightly lower voltage. Pete's version runs at about 340 volts.
A second 300 volt supply is built with a similar 230 to 250 volt transformer, a bridge rectifier and a large electrolytic capacitor, serious filtering not required. The negative end of this supply is connected to the positive terminal of the main capacitor in the raw B+ supply on the board. Do NOT connect this to the filtered supply on the board.....it WILL blow up the on board mosfets.
The positive end of this new supply goes to the Center Tap (usually red lead) of both output transformers. No other connections are made to this supply.
The lower supply IS the power supply built into the board with upgraded diodes to handle the extra current. With a power transformer in the 230 to 250 volt range the raw B+ will be around 300 volts. This enough to feed the on board screen regulator at 150 volts, and the filtered B+ which will be 5 to 10% lower than the raw B+. Everything except for the plates of the output tubes amp runs off of this supply as it did in Pete's version, but at a slightly lower voltage. Pete's version runs at about 340 volts.
A second 300 volt supply is built with a similar 230 to 250 volt transformer, a bridge rectifier and a large electrolytic capacitor, serious filtering not required. The negative end of this supply is connected to the positive terminal of the main capacitor in the raw B+ supply on the board. Do NOT connect this to the filtered supply on the board.....it WILL blow up the on board mosfets.
The positive end of this new supply goes to the Center Tap (usually red lead) of both output transformers. No other connections are made to this supply.
Thanks a lot George, now I got it. One last question, what are the VA rating for these transformers? Should be one (2nd?) much larger than another? I did quick search found antek has 200VA 230/6.3 toroidal I wonder 2 would be adequate.
You may want to look at some other alternatives to that 6GU5. I was curious what its curves look like, since the datasheet does not give them, and it is listed as a remote cutoff pentode. So here are 6GU5, 6CB6, and 6JC6 curves for comparisons. (6JC6 is 9 pin however, but available for $1 at ESRC)
All are 2 mA/div Vert. except 2), 50 V/div. Horiz.
1) 6GU5 pentode 0.2 V steps, 159 Vg2
2) 6GU5 triode mode 0.2 V steps (5mA/div Vert.)
3) 6CB6 pentode 0.3 V steps, 130 Vg2
4) 6CB6 triode mode 0.6V steps
5) 6JC6 pentode 0.15V steps, 92 Vg2
6) 6JC6 triode mode 0.5 V steps
Oops, fixup 6GU5 at same 2mA/div, as the other curves, for apples to apples compare:
7) 6GU5 triode mode 0.2V steps 2 mA/div Vert.
The remote cutoff is very obvious in the 6GU5 curves.
2 V peak to peak (ie, 1 V audio input) would be exercising the full span shown for the 6GU5 pentode curves. Connecting up the global N Fdbk however would cause the two 6GU5 grids to track each other better. (reducing the gain)
All are 2 mA/div Vert. except 2), 50 V/div. Horiz.
1) 6GU5 pentode 0.2 V steps, 159 Vg2
2) 6GU5 triode mode 0.2 V steps (5mA/div Vert.)
3) 6CB6 pentode 0.3 V steps, 130 Vg2
4) 6CB6 triode mode 0.6V steps
5) 6JC6 pentode 0.15V steps, 92 Vg2
6) 6JC6 triode mode 0.5 V steps
Oops, fixup 6GU5 at same 2mA/div, as the other curves, for apples to apples compare:
7) 6GU5 triode mode 0.2V steps 2 mA/div Vert.
The remote cutoff is very obvious in the 6GU5 curves.
2 V peak to peak (ie, 1 V audio input) would be exercising the full span shown for the 6GU5 pentode curves. Connecting up the global N Fdbk however would cause the two 6GU5 grids to track each other better. (reducing the gain)
Attachments
Last edited:
On the other hand, the global feedback won't help the 6GU5 any. For the same output power, the difference in the 6GU5 g1 swings must be the same.
I would even go so far as to use a 12GN7 or 12HL7 pair for the input differential stage (all 9 pin however), which I see got used for Pete's later P-P driver board.
12HL7 triode curves shown below. Notice the constant spacing, without "roll over". Best indication of grid1 not being too close to the cathode. (so good linearity)
On the other hand again, the 6CB6 was being used for (2nd H) distortion cancellation on each P-P side in the original design. (which would otherwise cause 3rd harmonic and higher odds in a P-P class aB amp) So the obvious 2nd harmonic in the 6CB6 was useful in that sense. (Umm, does that work though with a differential 2nd H cancelling front end? Maybe the 6CB6s are making cancelling 3rd harmonic in the differential stage.)
The 12GN7 or 12HL7 front end have so much more gain available at higher tail current, that they could pretty much clean up the output stage alone with the global Fdbk enabled.
I would even go so far as to use a 12GN7 or 12HL7 pair for the input differential stage (all 9 pin however), which I see got used for Pete's later P-P driver board.
12HL7 triode curves shown below. Notice the constant spacing, without "roll over". Best indication of grid1 not being too close to the cathode. (so good linearity)
On the other hand again, the 6CB6 was being used for (2nd H) distortion cancellation on each P-P side in the original design. (which would otherwise cause 3rd harmonic and higher odds in a P-P class aB amp) So the obvious 2nd harmonic in the 6CB6 was useful in that sense. (Umm, does that work though with a differential 2nd H cancelling front end? Maybe the 6CB6s are making cancelling 3rd harmonic in the differential stage.)
The 12GN7 or 12HL7 front end have so much more gain available at higher tail current, that they could pretty much clean up the output stage alone with the global Fdbk enabled.
Attachments
Last edited:
Back when the red board thread was active I collected just about every 7 pin tube that was pin compatible with the board and tested them for gain and distortion in the board with 6HJ5's as outputs. The 6HJ5's require more drive than the small sweep tubes that Pete originally used, and the usual suspects 6CB6, 6AU6, 6AG5, 6AN5, 6AH6, 6EW6, 6DC6, 6DK6, 6BE6..... fall short in the gain department. It requires several volts of drive to push the 6HJ5's to clipping.
I started looking at ways to get more gain that required board modifications, then I stumbled upon the 6GU5 which is a "Beam Hexode" with a Gm of 15,000 and pin compatible with the 6CB6. In tests on two red boards it consistently provided lower distortion in the critical 1 to 10 watt range than the 6CB6, and had enough gain to hit clipping with about 2 volts of input, if memory serves me correctly.
I couldn't find anything else that worked in the red board without wiring in a 9 pin socket, in which case there are several choices, but the overall result wasn't much better than the stock board with 6GU5's (similar distortion, more gain)...My red board just hits clipping with a CD player plodded directly into the inputs. With 95 db speakers, it is deadly!
At least a dozen boards were built with this combination, and the builders liked the results.
Another strange find that came from this testing....a 6BE6 pentagrid converter tube makes a real good triode, just wire ALL the grids in parallel.
I started looking at ways to get more gain that required board modifications, then I stumbled upon the 6GU5 which is a "Beam Hexode" with a Gm of 15,000 and pin compatible with the 6CB6. In tests on two red boards it consistently provided lower distortion in the critical 1 to 10 watt range than the 6CB6, and had enough gain to hit clipping with about 2 volts of input, if memory serves me correctly.
I couldn't find anything else that worked in the red board without wiring in a 9 pin socket, in which case there are several choices, but the overall result wasn't much better than the stock board with 6GU5's (similar distortion, more gain)...My red board just hits clipping with a CD player plodded directly into the inputs. With 95 db speakers, it is deadly!
At least a dozen boards were built with this combination, and the builders liked the results.
Another strange find that came from this testing....a 6BE6 pentagrid converter tube makes a real good triode, just wire ALL the grids in parallel.
Looking through the 7CM base tubes (6CB6 relatives) I found the 6EW6, which has high gm (14000 at 10 mA), sharp cutoff, and I had one here to test. So here are some curves comparing 6GU5 with 6EW6.
http://frank.pocnet.net/sheets/093/6/6EW6.pdf
All curves are 2 mA/div Vert., 50V/div Horiz.
1) 6GU5 pentode 0.2 V steps, 150 Vg2
2) 6GU5 pentode 0.4 V steps, 150 Vg2 (woops, looks like the 0V was off here, should reach same max current level as 1) )
3) 6GU5 triode mode 0.4 V steps
4) 6EW6 pentode 0.2 V steps, 125 Vg2
5) 6EW6 triode mode 0.4 V steps
6) 6EW6 triode mode ( probably near 0.4 V steps, 5 mA/div Vert. , I had this on file already, but no specs recorded)
7) and 6CB6 again in pentode 0.3 V steps 130 Vg2
So 6EW6 would be plug-in compatible with the same high gm of the 6GU5, but without the remote cutoff "feature" of the 6GU5. I had to reduce the Vg2 to 125 V to keep the 6EW6 pentode curves fitting on the same tracer scale as used for the 6GU5, so it is showing higher gm than the 6GU5.
( George, you must have had a bad 6EW6 back then, this one blows away the 6GU5 for gain)
There is a 6JK6 tube that is a sharp cutoff frame grid tube, 18000 gm, that fits the socket too. Listed at $18 though. And it is weirdly listed as 6JK6/6JL6 at both ESRC and VacuumTubes.net, which is NOT good, since 6JL6 is semi-remote cutoff. (not the same tubes at all!!)
If one could get a REAL 6JK6, I would recommend that. I don't have one to test unfortunately.
http://frank.pocnet.net/sheets/201/6/6JK6.pdf
http://frank.pocnet.net/sheets/093/6/6EW6.pdf
All curves are 2 mA/div Vert., 50V/div Horiz.
1) 6GU5 pentode 0.2 V steps, 150 Vg2
2) 6GU5 pentode 0.4 V steps, 150 Vg2 (woops, looks like the 0V was off here, should reach same max current level as 1) )
3) 6GU5 triode mode 0.4 V steps
4) 6EW6 pentode 0.2 V steps, 125 Vg2
5) 6EW6 triode mode 0.4 V steps
6) 6EW6 triode mode ( probably near 0.4 V steps, 5 mA/div Vert. , I had this on file already, but no specs recorded)
7) and 6CB6 again in pentode 0.3 V steps 130 Vg2
So 6EW6 would be plug-in compatible with the same high gm of the 6GU5, but without the remote cutoff "feature" of the 6GU5. I had to reduce the Vg2 to 125 V to keep the 6EW6 pentode curves fitting on the same tracer scale as used for the 6GU5, so it is showing higher gm than the 6GU5.
( George, you must have had a bad 6EW6 back then, this one blows away the 6GU5 for gain)
There is a 6JK6 tube that is a sharp cutoff frame grid tube, 18000 gm, that fits the socket too. Listed at $18 though. And it is weirdly listed as 6JK6/6JL6 at both ESRC and VacuumTubes.net, which is NOT good, since 6JL6 is semi-remote cutoff. (not the same tubes at all!!)
If one could get a REAL 6JK6, I would recommend that. I don't have one to test unfortunately.
http://frank.pocnet.net/sheets/201/6/6JK6.pdf
Attachments
Last edited:
Here are some more 6EW6 comparisons. I tuned the parameters (g1 steps or Vg2) to match up closely with the 6CB6 or 6GU5. Easy to compare gm now.
All are 2 mA/div Vert., 50V/div Horiz.
1) 6EW6 pentode 0.14 V steps, 132 Vg2
2) 6CB6 pentode 0.3 V steps, 130 Vg2
3) 6EW6 pentode 0.2 V steps, 150 Vg2 (top curve is at -0.23 Vg1 to fit it on the same 2 mA/div vertical scale)
4) 6GU5 pentode 0.2 V steps, 150 Vg2
So 6EW6 has 0.3/0.14 = 2.14 X the gm of the 6CB6.
And the 6EW6 is near matching (at -0.23Vg1) the gm of the 6GU5 at 0Vg1. (top curves, current increments about the same down to next lower curves, both at +150 Vg2) The 6EW6 would actually be somewhat higher gm, since its real (0 Vg1) curve is up another 0.23Vg1, off the scale. (about 2 graticule divisions up further)
So I would recommend trying out a "good" 6EW6 on these boards too. That was an RCA 6EW6 I tested here. (and Sylvania 6GU5, and RCA 6CB6A)
All are 2 mA/div Vert., 50V/div Horiz.
1) 6EW6 pentode 0.14 V steps, 132 Vg2
2) 6CB6 pentode 0.3 V steps, 130 Vg2
3) 6EW6 pentode 0.2 V steps, 150 Vg2 (top curve is at -0.23 Vg1 to fit it on the same 2 mA/div vertical scale)
4) 6GU5 pentode 0.2 V steps, 150 Vg2
So 6EW6 has 0.3/0.14 = 2.14 X the gm of the 6CB6.
And the 6EW6 is near matching (at -0.23Vg1) the gm of the 6GU5 at 0Vg1. (top curves, current increments about the same down to next lower curves, both at +150 Vg2) The 6EW6 would actually be somewhat higher gm, since its real (0 Vg1) curve is up another 0.23Vg1, off the scale. (about 2 graticule divisions up further)
So I would recommend trying out a "good" 6EW6 on these boards too. That was an RCA 6EW6 I tested here. (and Sylvania 6GU5, and RCA 6CB6A)
Attachments
Last edited:
I had 4 or 5 6EW6's, all used. The 6GU5's were NOS, so it is possible that 6EW6's were all weak in roughly the same amounts, but there is at least one other factor. I plugged all of these tubes into a working board with 6HJ5 output tubes I it and 220K ohm "Schade" feedback resistors from output plate to driver plate. No changes were made. I had found exactly 5 220K resistors rated for 800 volts, 4 of which were in the board, and still are despite eating roughly 1300 volts. The 6EW6 worked, so did the 6GU5. Different Rp's could account for the gain and distortion differences, by changing the feedback factor. From what I remember at the time they were roughly the same, but the 6GU5 was far cheaper. The price comparison may not be valid today.
I haven't looked further into the red board since I left Florida. My big red board amp, and all my other amps are still packed up. I am currently using a $25 Lepai amp. In my current concrete walled and floored basement, it doesn't matter....the acoustics suck!
I have designed a new driver board and the first prototype is being debugged while finding the glow point for some of our favorite Korean 13GB5's. The pair of tubes in the amp last week worked far better than expected.
The new driver board uses 9 pin tubes for two reasons. There are far more pin compatible 7 pin tubes to choose from but few with high Gm. There are several good choices for cheap high Gm tubes in the various 9 pin flavors. Also, there are two non compatible (different pin circle) flavors of 7 pin sockets, and the good ones are hard to find. 9 pin sockets are far more common.
Good point about the Rp's being different. I see 6CB6 lists 280K for Rp, 6EW6 lists 200K, and 6GU5 lists 165K. One can see some difference in the slope of the plate curves above too.
I see 6CB6 listing at $3 now. 6GU5 at $3 too. And 6EW6 at $4 at ESRC. Beats the $40 on Ebay for the 6JK6. ($18 from ESRC) 6JC6 still on the $1 list at ESRC. 12HL7 on the $1 list too. Maybe I should sell some 7 pin socket adapters for 6JC6 to replace 6JK6!!
I see 6CB6 listing at $3 now. 6GU5 at $3 too. And 6EW6 at $4 at ESRC. Beats the $40 on Ebay for the 6JK6. ($18 from ESRC) 6JC6 still on the $1 list at ESRC. 12HL7 on the $1 list too. Maybe I should sell some 7 pin socket adapters for 6JC6 to replace 6JK6!!
Last edited:
I have my red board in a box still. Never finished it, although I did solder all the parts. Only need to buy the iron and create a case for it. Nice to see some new activity here, it prompts me to do something with my pile of stuff.
As far as first stage tubes are concerned I bought a bunch of 6DK6's at the time. Any thoughts on those?
As far as first stage tubes are concerned I bought a bunch of 6DK6's at the time. Any thoughts on those?
First time posting to this thread with my first time build. I've wanted a tube amp ever since I listened to a Cary Audio amp. However Cary Audio is way out of my price range and I was turned off by the so-so reviews of the tube amps coming out of China. I also want to keep learning new things so I figured why not do something I've never done before, like build an amp? Although I've been in international broadcasting for 30 years I am in the program format side and not the technical side. However a friend of mine who is more technical offered to mentor me along the way. So I chose the DCPP amp. If I had known 3 months ago what I know now, I probably would have chosen something a bit more simple. But it has been a learning experience. Below are pictures of my build so far. It is a stock build right down to the chassis plate. As you can see the switch has different dimensions than the specified cutout. that was the only discrepancy on the plate design. The first time we turned it on the tubes glowed. Unfortunately after about 4-5 minutes we had smoke come from the R21 resistor. We shut it down, resoldered R21 and while we didn't get smoke the measured temperature was at 230 Fahrenheit and climbing when we turned it off. An email to Pete Millett revealed my noobie mistake was not attaching the mosfets to the chassis plate. I've ordered two more mosfets and R21 and will try again. It has been fun and a learning experience.
Attachments
I agree Dagwood. Pete has been more than helpful and I appreciate his patience with me. I will reiterate that I would have never attempted this without the help of a friend who has years of electronics experience . He has taught me a lot including letting me make some mistakes along the way that I had to go back and fix with his explanation of why it was a mistake. I have another friend who is an E.E. with a lot of experience in high power transmitters. He has come in later in the process, looked at my board and has made some recommendations that should lead to an ultimately successful project completion. To those who want to make a start in this great hobby but like me have no experience in electronics I would highly recommend finding somebody willing to help through the process who can also explain what and why the different components are important.