I have been avoiding measurements as it will take time away from listening to it long enough to really take in its characteristics. (Me slow learner!) The one thing that I noticed right from first power-up was a slight covered or muffled sound. The amp was first the sole speaker driver and then set up as the driver for woofer only (biamped system) . The sound as though the speaker has a woolly blanket on it persists.
I'll get a couple of other tube types to try in order to see what the effect of the tubes alone is. The results there will decide what to test for.
Otherwise the circuit is very quiet (during power up and down as well) and while the output stage currents can vary by as much as 15mA from session to session, both channels are doing so together so I take it that mains voltage and room temp variations are the chief driver of that, (though I have wondered a couple of times if the covered sound could be caused by oscillation and if so, perhaps that's what nets the current changes.
Also, I saw Francois asked about output transformers. It started with 5K R-cores but I thought the muffled sound might be related to them so switched to the only other ones at hand, 3K. They sound a lot better. Not sure how much of that is impedance value and how much is manufacture. (The soft sounding 5K were Softone.)
With the 2K resistor the screen voltage starts to drop at 11W. Without the resistor the screen voltage stays solid at 190V regardless of output level.
Last edited:
I have a few 6DQ5 tubes with no partner so decided tonight to try and crank the bias up on one and see what happens to the measurements and/or tube. Well with my DMM showing 110mV or 110mA of bias I get 0.5% THD at 1 watt and 5% THD at 13.7 watts. I did not notice any red plating but the pull cord on my overhead shop light broke so I can't turn it off to check under low light.
This being the measurements I was expecting except at a bias around 80mA I would like to do a sanity check. What is an alternative method to taking a voltage reading off R117 and R217 for measuring bias at the output tubes?
This being the measurements I was expecting except at a bias around 80mA I would like to do a sanity check. What is an alternative method to taking a voltage reading off R117 and R217 for measuring bias at the output tubes?
0.5% at 1 watt and 13.7 watts at 5% look better. But your 6DQ5 is dissipating 34 watts! Not the efficiency expected from UNSET topology.
I revisited an old Glass Audio article on a “conventional” SE 6DQ5 implementation, and the author claimed “much less than 1% distortion”, and 14 watts output at 24 watt plate dissipation. The operating conditions and loading are interesting. See the attachment.
Can’t think another way to measure bias current, except for inserting a milliamp meter in place of R117, R217.
Attachments
Perhaps my putting it in question form made it too vague.
To know the quiescent current without measuring across the 1 Ohm resistor you can measure the DCR of your output transformer primary. Then, measure the voltage at DC across it while it's running. Knowing the DC resistance of the primary you can use Ohm's law to calculate the current.
Or, if you want to make it so you don't have to calculate each time, just put another 1 Ohm resistor in series with the output transformer primary (At OPT1, for example ) and measure across that.
To know the quiescent current without measuring across the 1 Ohm resistor you can measure the DCR of your output transformer primary. Then, measure the voltage at DC across it while it's running. Knowing the DC resistance of the primary you can use Ohm's law to calculate the current.
Or, if you want to make it so you don't have to calculate each time, just put another 1 Ohm resistor in series with the output transformer primary (At OPT1, for example ) and measure across that.
Thanks Hearinspace. I was able to verify using that method to double check myself.
Still no measurements but have been playing the amp about an hour a day in the background while doing other stuff.
Started with 6jc6 > 6CD6 and noticed a kind of closed in , blanket over the microphone/singing into a bucket full of cotton kind of sound in low mids. Swapped out the 6CD6 for a 6DN6 and there was improvement, a bit more gain and some added oomph. The 6DN6's themselves had some kind of coating on them and filled the room with burning plastic smell so took them back out and scrubbed with Comet. While wet put back the 6cd6 and instead changed the input tube to 6ha6/6hb6. Covered over sound gone.Nice and clear. Maybe the 6jc6 was oscillating? Don't know but anyway much better.
Then 6DN6 back in yesterday and no smell and much better sound. (6ha6>6dn6 - 6ha6 Vp≈127 /6DN6 Ip≈55mA.)
In the interim there somewhere plugged in 6BG6 but both tubes no conduction. Probably needs higher Vg2. and maybe different R values to get it. Don't know but at the time didn't want to risk frying something and having to bench the amp. Will look at the data sheet later.
Anyway, first conclusion - This is a neat little amp with lots of flexibility and potential for different uses - and conversion to parallel mono or PP.
Also switching the few tubes tried it's clear the output sound can be shaped a lot with use of different tubes and combinations of tubes just plug and play. xpect better caps will be audible too.
Lots of promise.
Started with 6jc6 > 6CD6 and noticed a kind of closed in , blanket over the microphone/singing into a bucket full of cotton kind of sound in low mids. Swapped out the 6CD6 for a 6DN6 and there was improvement, a bit more gain and some added oomph. The 6DN6's themselves had some kind of coating on them and filled the room with burning plastic smell so took them back out and scrubbed with Comet. While wet put back the 6cd6 and instead changed the input tube to 6ha6/6hb6. Covered over sound gone.Nice and clear. Maybe the 6jc6 was oscillating? Don't know but anyway much better.
Then 6DN6 back in yesterday and no smell and much better sound. (6ha6>6dn6 - 6ha6 Vp≈127 /6DN6 Ip≈55mA.)
In the interim there somewhere plugged in 6BG6 but both tubes no conduction. Probably needs higher Vg2. and maybe different R values to get it. Don't know but at the time didn't want to risk frying something and having to bench the amp. Will look at the data sheet later.
Anyway, first conclusion - This is a neat little amp with lots of flexibility and potential for different uses - and conversion to parallel mono or PP.
Also switching the few tubes tried it's clear the output sound can be shaped a lot with use of different tubes and combinations of tubes just plug and play. xpect better caps will be audible too.
Lots of promise.
The author describes the OPT in that article as a push pull transformer - i.e. no air gap, and the output power and distortion claims look dubious to impossible to me. I'm not sure how you eyeball less than 1% distortion from a scope trace.
Maybe I'm just tired and misread it - it's pretty late here.
Many here know that I spent the first 62 years of my life in South Florida, and much of my spare time was spent in the Atlantic Ocean on a boat of some kind. This has resulted in multiple incidences of skin cancer and the associated treatment or surgery required to stay ahead of it.
Early this year I had a tumor removed that had grown too deep due to Covid delays. After a two hour session in the OR under general anesthesia the local general surgeon stated that he would not perform any more surgery on me and that I needed to be referred to a specialist.
Due to the craziness in the US health care and insurance system, this visit kept getting delayed. I actually chopped off a minor tumor myself with a pair of wire cutters during the 8 months I was on the wait list.
So late last month I get to see the specialist, who of course finds all sorts of stuff that needs "immediate attention" which must be "completed before the end of the year." Three visits later I find myself scheduled for three surgeries in a two week period. I now have a good sized hole in my forehead, a bigger hole in my left shoulder, and a third surgery scheduled for Thursday. Judging from the lump on my right shoulder, it will be the biggest hole of the three.
The hole in the head required the use of a Hyfrecator 2000 on my forehead. This is a 0 to 35 watt (set on 12) radio transmitter on 450 KHz that turns cancer into charcoal dust. It however left me with a monster headache that lasted for 4 days, some blurry vision, and general stupidity. Most of those effects are now gone.
This needs to be done, but has left me incapable of finishing the workbench, or pretty much anything else physical. My left arm is immobilized, and my right arm will probably be also after Thursday.
The UNSET board is back on the bench, but it hasn't seen power in a while, and none of my test equipment is connected up. The exact sequence of events in the next few weeks is pretty much beyond my control.
There is a known variation of current draw with line voltage. This is a result from having the bias and screen voltages regulated, but the plate voltage moves with the line voltage. There is a way to fix this, but it involves flying wires and a resistor. Each tube / supply voltage combination is different, so it's something that gets done after the builder decides which tubes go in the final amp build.
The resistors on either ends of the pots (R9 and R111 / R211) can be tweaked to center the pot and adjust it's range. For now, the published values allow for a very wide range of bias voltages to accommodate everything from a 6W6 to a 6LW6 over supply voltages from about 300 volts to 650 volts. Once the choice of output tube and supply voltage is known, the resistors can be chosen to make the pots far less twitchy. If this ever gets to a new layout stage, I can dual footprint the pot so a 10-turn pot could be used.
The current in the output tubes seem to ramp up faster than the driver tubes. This is highly dependent on tube choice. As the driver tube warms up it's plate voltage drops which tends to force the output tube's current upwards depending on the value of the coupling cap. With a 1 meg input impedance to the output stage, large value coupling caps are not needed or wanted.
Much of my testing has been done with regulated bench supplies so that I can try different voltages easily and change then quickly. I did have a board running from one of three different power transformers several times over the last year. I never saw any motorboating or heard any weird stuff on power up, but some more testing is needed. As I stated previously, I need to duplicate the exact setup that some of you are using to see if I can duplicate these issues.
I use the resistor in the mosfet drain supply trick in several places where I desire to limit the maximum current.
Here it is in the screen grid supply where it serves two purposes, current limiting, and mosfet SOA protection with high B+ voltages. As the screen current rises the voltage drop across the resistor increases. The mosfet will hold its source constant within a few tenths of a volt until the drain voltage drops to a value near the source voltage. At this point the voltage drops like a rock in a shallow pond. The resistor value should be chosen such that the screen voltage stays constant until maximum dissipation is approached, then it is allowed to drop. I often include a capacitor in the 10 uF range from the mosfet's drain to ground to allow for a brief overload during a musical transient.
Every tube is different, and in some cases two tubes with the same number and manufacturer will draw grossly different amounts of screen current depending on how well the grids were aligned when the tube was made.
In the UNSET design the cathode bias is derived from the screen supply. Allowing the screen voltage to collapse except when absolutely necessary can lead to instability, so this resistor should be as small as possible, and may not be needed at all at more reasonable B+ voltages than the 550 to 650 volts that I was using when I kept blowing parts.
The charts that I published were all made with a 26HU5 tube. I used this monster tube (similar to the 26LW6) so I could push the dissipation up to about 50 watts without blowing anything. The 26HU5 and 26LW6 have plates that are about 50% larger than the 6DQ5, so they would produce a bit more power than the 6DQ5. Exactly what the difference is can only be determined by trying both tubes in the same amp.
The theoretical maximum efficiency attained in a class A amp is 50%, and approaching that number usually requires class A2. Since some of my numbers exceed 50% by a percent or two, there are obviously some errors present. These numbers are highly related to losses in the tube, so a smaller tube (the 6DQ5) will always present a lower efficiency than a larger 26HU5 under identical conditions. The DCR in my 1500 ohm OPT (run at 3000 ohms) is another place where a percent or two is gained.
I published those charts to provide some insight as to what OPT impedance to use with a given B+ and maximum tube dissipation. They should not be taken as an absolute predictor of power output since I used a big tube and a low loss OPT. If I had used my Transcendars, the numbers would be considerably lower.
That article smelled funny so I did some simple stupid math. The B+ voltage is 200 volts. The load impedance is 1622 ohms. Lets just say that the author has a theoretically perfect tube that can put all of the power supply power into the OPT, and the OPT is also theoretically perfect. If the tube were to fully conduct, there would be zero volts on the plate, so the plate end of the OPT is at zero volts. On the next half cycle the tube is fully off (non conducting). Our perfect OPT will "fly" the plate end up to 400 volts. This puts the MAXIMUM Peak to Peak voltage across the OPT at 400 volts. 400 Volts P-P is about 142 volts RMS. 142 volts RMS across 1622 ohms is 12.43 watts, so I call BS on the entire article! In reality he was probably in the 10 watt range if it even worked at all.
The OPT was intended for a push pull amp. My experiments have taught me
that I might get 2 or 3 watts in SE through a P-P OPT that's rated for 80 watts. That was running a 6V6GT on 30 mA. Attempting to get 15 watts from a 6DQ5 is going to take a lot of current on a 200 volt supply, but we already know that it is impossible. My 1500 ohm OPT graphs start at 300 volts and I still didn't get 15 watts.
The 6BG6 is a 6L6GA in disguise. It needs 300 Volts or more on the screen grid to do anything.
Yes, the thing is essentially a tiny Tesla coil. The end that produces a fireball is on a probe that zaps the patient.
https://www.conmed.com/en/products/...it/the-hyfrecator-2000-electrosurgical-system
Thanks for your replies George. All good on the Unset board, even the apparent motorboating sorted itself out fairly quickly and the best I can figure is that it was either a fault of one of the first tried tubes or maybe a poor pin to socket connection as there's no trace of trouble now.
As noted in the post above (copied from my notes) , the amp is working very well and its adjustability is a real plus ie. it allows room for fine tuning to match the other amp I'm using in a bi-amped setup. There are no negatives to complain about.
Glad your nose is back online.
Forgive a serious comment: I don't like to think you might be laid low again after Thursday's rf treatment but what's important is how you are and not what you will do on the board for us. Best would be to know you are taking whatever time suits you for an easy going reboot.
Thanks!
As noted in the post above (copied from my notes) , the amp is working very well and its adjustability is a real plus ie. it allows room for fine tuning to match the other amp I'm using in a bi-amped setup. There are no negatives to complain about.
Glad your nose is back online.
Forgive a serious comment: I don't like to think you might be laid low again after Thursday's rf treatment but what's important is how you are and not what you will do on the board for us. Best would be to know you are taking whatever time suits you for an easy going reboot.
Thanks!
So far, the only negative thing from the RF zapping today and Monday, was the smell of crispy fried George. It's somewhat similar to the smell of a burnt mosfet. Last week they put the probe on my forehead. That made my brain hurt. Today they fried up my right shoulder. I didn't feel anything, but they had injected lots of nerve deadener. So currently I have stitches in both shoulders and am not supposed to lift anything or drive a car. I can work the mouse OK though.
Seriously, I have been getting these kinds of treatments for nearly 30 years. 62 years in the Florida sun will do that to a fair skinned blonde. As long as I stay ahead of the growths, I should be OK.
I'm still trying to get all my test equipment hooked back up on the workbench. The PC with REW that I built a month ago is not cooperating right now either. Might take a few more days. I seem to have a total of eight 6DQ5's in boxes. No two are exactly the same. I know that I have a box full of used ones somewhere from a lifetime of collecting them. Since they seem to be the tube of choice right now, I'll try them first.