Inverted OTL with 6SN7/6AS7

What people are working on at the moment
Post Reply
Wolfgang
Old Hand
Posts: 229
Joined: Mon May 07, 2018 3:08 pm

#61 Re: Inverted OTL with 6SN7/6AS7

Post by Wolfgang »

Because I started this thread I feel kind of responsible and didn't want to miss my chance to post the updated and final schematics.

I also found the cause for the slight down slope of the frequency response. It's the missing bypass cap parallel to the first cathode resistor. I didn't want the input to be so extremely sensitive (0,6Vp-p with cap) so I decided from the beginning not to use this cap. With it the amp has a ruler flat frequency response! I didn't test how it sounds with the cap because I am absolutely satisfied as is and don't want to bring down the signal level by using the cap together with some voltage divider at the input. But it's an option if one wants ruler flat frequency response.

I added grid stoppers because with one set of Russian made 6SN7 I got some little oscillation tendencies. Now this is also fixed. No deterioration of sound with the grid stoppers. Rather some added clarity.

I also had a bad tube that sparked over from anode to grid which made a nice fat "popping" sound (no damages). The TVS diode will hopefully stop this if it happens again with another tube. But I will also buy better tubes from now on.

The DC offset can be controlled by using 1%/20ppm Vishay resistors and match pos and neg side so that the amount of heat dissipation per resistor is the same. I use 16 40ohm/10W( parallel/series) for the pos rail and four 7ohm/10W (parallel/series)for the neg rail. Like this each resistor dissipates about 1,5W. The schematics still show my original setup with typical 5%/250ppm resistors. Adjustment is easy: after 6,5min(4M resistor!) the relays open and with a reading of about +200mV (without speakers connected) the DC offset will stabilize at around -50mV after about 20 min. This will also depend on the tubes of course. One can get it down to -10mV by starting with a higher pos value. The Maida regulators add very much to the stability of the DC offset. With a typical CLCLC there is some fluctuation.

After listening for some weeks I can only say that it is the best sounding amp I have ever built and by far the best sounding OTL.
on_off control circuit.png
InvOTL.png
User avatar
Paul Barker
Social Sevices have been notified
Posts: 8863
Joined: Mon May 21, 2007 9:42 pm

#62 Re: Inverted OTL with 6SN7/6AS7

Post by Paul Barker »

Interesting
"Two things are infinite, the universe and human stupidity, and I am not yet completely sure about the universe." – Albert Einstein
User avatar
Ray P
No idea why I do this anymore
Posts: 6294
Joined: Thu Nov 22, 2007 5:18 pm
Location: Somerset

#63 Re: Inverted OTL with 6SN7/6AS7

Post by Ray P »

Wolfgang wrote: Fri Aug 02, 2019 7:33 pmAfter listening for some weeks I can only say that it is the best sounding amp I have ever built and by far the best sounding OTL.
You keep tempting me....
Wolfgang wrote: Fri Aug 02, 2019 7:33 pmThe DC offset can be controlled by using 1%/20ppm Vishay resistors and match pos and neg side so that the amount of heat dissipation per resistor is the same.
I know you're thinking about regulated supplies for the output stage power supplies but have you considered using a capacitance multiplier in the power supplies for the output stage, perhaps as a halfway house - you'll get low ripple and lots of current and it might be better with respect to what look like thermal issues messing with the DC offset?

I recently bought some boards from this group buy for my MoFo project (one day!);

https://www.diyaudio.com/forums/group-b ... ly-gb.html

There's a dual rail board that, IIRC, will run with assymetric voltages as each half is a full circuit with no shared parts (other than the ground rail).
Sorry, I couldn't resist!
Wolfgang
Old Hand
Posts: 229
Joined: Mon May 07, 2018 3:08 pm

#64 Re: Inverted OTL with 6SN7/6AS7

Post by Wolfgang »

I will try this, Ray. Thanks! It looks to be ideal for this job considering amps,voltages, heat if I can make it work with the resistors/(smaller value) cap/chokes. Do you know the max input voltage for these boards? I simply couldn't find this information.

The thermal issues/DC offset fluctuations are no longer a problem but the power resistors will have to stay the same no matter what psu I will use because I need the voltage drop for the zero DC and the idle current. The SLB will only give me clean, more stable power with almost no ripple current.

My main goal from the beginning was to get the psu noise down to the lowest possible level with this amp. If it works as good as the Maida regulators it will improve the performance even more. The amp has zero hum but there is some minor "electric noise" that comes from the CRC psu.

The "thermal issues" are not even so hard to grasp -although it took me a little while to figure it out because I was searching in the wrong direction. The resistors change their resistance a little depending on the heat which they have to dissipate(that's why the low ppm/1% is so important) and as the voltage drop is different at the pos and neg rail it will effect the DC offset if the resistors change their values differently . It is like using a little different than correct values for both rails. Add to it the tubes (12 triodes!) which don't conduct exactly the same current but naturally fluctuate a bit and DC offset can be really OFF if no counter measures are taken..
User avatar
Ray P
No idea why I do this anymore
Posts: 6294
Joined: Thu Nov 22, 2007 5:18 pm
Location: Somerset

#65 Re: Inverted OTL with 6SN7/6AS7

Post by Ray P »

I think the max input voltage is 51VAC, dictated by the specs of the chip running the active rectifier - did you notice the SLB uses them. Here's a link to a sperate thread describing the active rectifier;

https://www.diyaudio.com/forums/group-b ... er-gb.html

On your Inverted OTL project, wouldn't it be good to minimise the thermal aspect of the DC offset by not having to burn off lots of volts? With a cap multiplier like the SLB you don't need to do that for ripple reduction. If you run separate 'right sized' transformers for each side of the output power supply you ought to achieve better thermal stability and still get very low ripple.
Sorry, I couldn't resist!
Wolfgang
Old Hand
Posts: 229
Joined: Mon May 07, 2018 3:08 pm

#66 Re: Inverted OTL with 6SN7/6AS7

Post by Wolfgang »

The SLB is really the best input for me at this moment and I have to study more details now in order to figure everything out.
On your Inverted OTL project, wouldn't it be good to minimise the thermal aspect of the DC offset by not having to burn off lots of volts? With a cap multiplier like the SLB you don't need to do that for ripple reduction. If you run separate 'right sized' transformers for each side of the output power supply you ought to achieve better thermal stability and still get very low ripple.
Sorry for the long answer but I couldn't make it shorter and explain all the factors.

The InvOTL does everything a little different - it isn't called InvOTL for nothing. The output psu needs to be asymmetrical in order to be able to deliver a symmetrical sine if the tubes are operated like this. The max peak of the undistorted sine (or music signal) depends on the idle current and the idle current changes proportionally with the asymmetry of the voltages of the psu. The neg bias for the anode is reflected in the neg rail voltage of the psu and appears as DC offset if bias and voltages of the psu aren't perfectly adjusted/matched.
If one would decrease the voltage differences between pos and neg rail (more voltage on the pos rail) the idle current would go up, the bias would need to be adjusted downwards and the neg rail voltage would only change very little. The tubes sound different if the idle current changes. This is not so much the case with tubes that are operated with typical grid bias. It also happens here with cold or hot biased tubes but the effect on the sound seems to be less.
I started with 1100mA and decreased it downwards -adjusting everything every time - until I found the best sounding (to my ears) operating point roughly between 720-800mA. In this range I could get the most satisfying, euphonic, saturated sound from bass to treble. That's how I ended up with these settings. But that's not everything.
If the asymmetrical voltages are not checked/adjusted with the help of an oscilloscope one could end up with settings that would make one side of the sine distort (soft clipping) already much earlier than the other. With these settings I get 16Vp-p into 15ohm with no side of the sine starting to soft clip earlier than the other. If I go up 0,1V from 1,1V p-p input signal one side shows first signs of clipping while the other side starts to flicker lightly. Another 0,1V and the other side follows. It was a long process to get to this point but it's the best I could find considering all of these factors I have just described and it sounds like it is pretty ideal. Dynamic jumps start without the slightest delay (no NFB) and dynamic signals are never compressed or distorted.
Only by adding an output cap one could make this process a lot easier because one would only have to adjust bias and current and the DC offset wouldn’t matter. But then again the problems with caps are “cunning” as described in the book “Current-driving of loudspeakers” by E.Merilaeinen. The harmonic distortions of caps are negligibly small but regarding SQ the dielectric hysteresis relating to their insulator material (similar to ferromagnetic hysteresis) is more essential. The higher the dissipation factor of the cap (tangent of the loss angle) the more there is room for hysteresis-derived delay between voltage and charge. And the “cunning” part is this: This usually doesn’t show up in sine wave measurements (except ceramic caps) but it can produce some detriment on real signals.
So, it's not because of the ripple rejection but because of the most satisfying sound that I have chosen these settings with these values for the resistors. And 800mA give still so much headroom even for the hardest dynamic jumps (with sensitive speakers) that this amp will never compress any signal or run out of power.
Wolfgang
Old Hand
Posts: 229
Joined: Mon May 07, 2018 3:08 pm

#67 Re: Inverted OTL with 6SN7/6AS7

Post by Wolfgang »

On your Inverted OTL project, wouldn't it be good to minimise the thermal aspect of the DC offset by not having to burn off lots of volts? With a cap multiplier like the SLB you don't need to do that for ripple reduction. If you run separate 'right sized' transformers for each side of the output power supply you ought to achieve better thermal stability and still get very low ripple.
I looked in my documentation of this project at what I did in the beginning (measurements, sound description). I took S.Bench's recommendation (see below) as a starting point but opted for a 60VCT for some headroom because I wasn't sure how many tubes I would end up using and at which idle current depending on how it would sound and how well it would be able to drive my speakers (regarding impedance) under real conditions.

"Hints on a 20 watt version using 6AS7s
You can get 20 watts SE OTL from 12 paralleled 6AS7s. The only thing that needs changed on the schematic is the power supply and the bias. The bias changes to -120 volts (-152 volts to account for the cathode voltage). The tubes need 32 volts at 170 mA per tube (85 mA per section), so going to a 48V CT transformer is appropriate. This will develop +/- 32 volts. The filaments are arranged across this transformer as 4 series on one side, 4 series on the other side and 4 more series on one of the sides. Drive required is (only) 240V p-p (= 170V rms), so you could also lower the 70k plate resistor on the 6CG7 to increase the slew capability. The output choke should probably become a 100 to 500 mH 3 (or so) Amp inductor. Fair Radio has some of these."

First I went of course with his recommendations (except the number of tubes, transformer voltage, higher inductance of the output choke) and I was there....at these bias voltages, idle currents. And of course the heat that needs to be burnt with resistors decreases a lot if the idle current and the voltage of the positive rail get higher. For the driver stage I went almost exactly with the recommendations (260V p-p instead of 240Vp-p).
But I didn't like the more "aggressive" sound compared to the softest pastel sound colors with a dream come true wide open sound stage at lower bias settings. And the lowest possible DC offset was higher with the higher settings settings/less heat burning with resistors. I was aiming for stable 0V DC (+/-10mV) which I have achieved.

The thermal aspect has been fully solved with the number/combination/quality of resistors and I get the sound quality which is possible with this amp. I think the thermal issue is not a bigger problem than the heat developed by 12 6AS7. It just adds to the power consumption of this amp.

For anybody who would like to build this amp there are almost limitless possibilities for experimenting with different settings/SQ. What I did and have chosen as my final settings is just one out of many ways how to do it.

BTW, I ordered boards, parts and transformers for the SLB experiment. Will report back when I have tested it. I have the highest expectations comparing the specs and how it operates with the needs of the InvOTL output stage.
Wolfgang
Old Hand
Posts: 229
Joined: Mon May 07, 2018 3:08 pm

#68 Re: Inverted OTL with 6SN7/6AS7

Post by Wolfgang »

I forgot to mention that the next lower value (using Avel toroids) would be at 50VCT which would lower DC per rail for about 8V. Don't think that would make much difference regarding thermal issues but would probably throw off the balance I have achieved with DC offset,idle current and sound.
User avatar
Ray P
No idea why I do this anymore
Posts: 6294
Joined: Thu Nov 22, 2007 5:18 pm
Location: Somerset

#69 Re: Inverted OTL with 6SN7/6AS7

Post by Ray P »

Hi Wolfgang. I think I understand the concepts behind the Inv-OTL output stage power supply and, on reflection, I'm not really sure the SLB will help much with your DC offset drift; it'll give you very low ripple, which may be advantageous, but while you're using stacks of resistors to get the right voltage/current, especially with a symmetrical CT transformer where one half of the PS has to burn a lot more volts than the other, surely there will still be thermal affects that will result in some DC offset drift. The SLB modules aren't regulated power supplies - maybe that's what's needed?

Some more questions...

Does this amp warrant custom wound transformers to optimise the performance?

Will you have a DC monitoring arrangement to disconnect the speakers in the event of it exceeding some threshold? If during normal extended running, as opposed to exploring the initial config parameters, DC offset increases can it be adjusted down with the preset or will you have to go back to tweaking the power supply resistors? If the former, it might be good to have the adjustment on the front or top panels, maybe with some voltmeters to display the DC offset?

I've been thinking about your logic circuit to turn things on/off in the right order. I wonder if the same functionality could be achieved with a simple rotary switch if you accept doing timings manually?
Sorry, I couldn't resist!
ratbagp
User
Posts: 30
Joined: Sat Mar 26, 2016 2:05 pm

#70 Re: Inverted OTL with 6SN7/6AS7

Post by ratbagp »

Perhaps a simpler solution for the DC offset turn on/off problem is to switch in a capacitor to protect the speaker at turn on and at turn off. When the amp is stabilized and you are listening, just switch the cap out of the connection to the speaker.

ray
User avatar
Ray P
No idea why I do this anymore
Posts: 6294
Joined: Thu Nov 22, 2007 5:18 pm
Location: Somerset

#71 Re: Inverted OTL with 6SN7/6AS7

Post by Ray P »

ratbagp wrote: Tue Aug 06, 2019 2:14 pm Perhaps a simpler solution for the DC offset turn on/off problem is to switch in a capacitor to protect the speaker at turn on and at turn off. When the amp is stabilized and you are listening, just switch the cap out of the connection to the speaker.
Good to see you here again Ray.

That would enable you to start listening sooner but it won't protect your speakers if high DC occurs during listening.
Sorry, I couldn't resist!
ratbagp
User
Posts: 30
Joined: Sat Mar 26, 2016 2:05 pm

#72 Re: Inverted OTL with 6SN7/6AS7

Post by ratbagp »

Good point. It's an intriguing amp.

ray
Wolfgang
Old Hand
Posts: 229
Joined: Mon May 07, 2018 3:08 pm

#73 Re: Inverted OTL with 6SN7/6AS7

Post by Wolfgang »

Ray,
I don't have any DC offset issues any more as I tried to explain already because of the better power resistors I use and how I use them.
I also don't need the SLB because of any voltage drift or DC offset issues. I need it only for lowest ripple current which should result in the darkest possible background for the music signal and rule out the last avoidable noise source for a non-NFB amp. A typical regulated psu wouldn't be helpful in this position as I had figured out over time. That's why I liked the idea of the SLB psu so much.

I cannot really answer the transformer question from a performance point of view. But I don't think it matters. But from a practical point of view it would reduce the number of transformers and space needed. I use 5 transformers: 2 for the driver stage (each with heater output for the 6SN7), 2 for the output psu, one for the heaters of the 6AS7. Relays and delay/sequence on/off circuits have separate psus.

The "DC monitoring" I simply solved with stable DC offset around +/-10mV. It starts at around +200mV when the speakers are connected (6,5min warm up time), settles at around -10mV after about 10-12min and stays there. No need for monitoring.
However, it's helpful if you can reach the DC offset adjustment pots easily because you can simply plug in a voltage meter in the speaker terminals and adjust/check the DC offset. It might change a bit with the aging tubes. DC offset is only controlled by the pot. The power resistors set the idle current.

I had originally planned to use a DC control circuit that would also disconnect the speakers if DC offset reaches a certain (adjustable) DC level. The problem here was the following: typical DC protection circuits are triggered only by relatively high DC (0,7V-1,5V) levels. The simple passive circuits (crowbar)are not good enough and will negatively effect the output signal and shouldn't be used with this amp IMO.

In the case of the InvOTL we have to deal with 3 possible causes but all different from the cause that triggers normally DC protection circuits.

• The anode bias voltage at the output in case of an eventual/possible tube discharging from anode to grid is taken care of by the TVS diode. It clamps the voltage at around 70V which is good enough for this purpose. Reaction time is some usec. Full physical tube short circuits are not covered really by this but would be in combination with a DC protection circuit that also would turn off the psu & speakers. A different TVS diode can be used if a lower psu voltage would be chosen with different voltage/idle current settings. In my version it has to stay closed up to 43V.
• DC offset is no longer an issue IMO. If DC offset of the InvOTL should be covered as well typical DC protection circuits are not sensitive enough. Because of this and when I still thought I would need one I developed a little circuit that triggers already starting at around +/-50mV and can be adjusted to higher levels. But I didn’t develop this further because I thought it wasn’t necessary anymore. This circuit could be used either to simply turn on a LED (per channel) as a reminder for necessary manual adjustment or trigger the relays automatically and also protect the speakers. It would have to be set of course a little higher than the highest DC offset when the amp is already connected to the speakers by the logic circuit but is still in the warm up phase before it reaches its lowest DC level. That would be a little above +200mV in my case.
• Full DC voltage of the output psu at the output until the tubes start conducting has been taken care of by the automatic switch.

Three simple switches would do the job. Power outage or a short interruption of the mains power is the only problem if it’s not controlled by some circuit. In this case you would get at least some DC at the output. How much would depend on the time of the power interruption before it comes back again or how much the tubes could still drain the caps before the heaters are down. In either case it would need a quick reaction time for the listener to manually turn off the amp/speaker relay before the power comes back on
User avatar
Ray P
No idea why I do this anymore
Posts: 6294
Joined: Thu Nov 22, 2007 5:18 pm
Location: Somerset

#74 Re: Inverted OTL with 6SN7/6AS7

Post by Ray P »

Sorry Wolfgang, I obviously misunderstood - I thought you were still having an issue with DC offset drift. In that case it will be interesting to see what the SLB delivers, or if their are any unexpected issues.
Sorry, I couldn't resist!
User avatar
Ray P
No idea why I do this anymore
Posts: 6294
Joined: Thu Nov 22, 2007 5:18 pm
Location: Somerset

#75 Re: Inverted OTL with 6SN7/6AS7

Post by Ray P »

Have you tried the SLB yet Wolfgang?

Ray
Sorry, I couldn't resist!
Post Reply