• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

When using a DC filament supply on a DHT...

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Well, this has pretty much been one big fat failure so far :( I got the regulated DC filaments working, but my noise/hum level at the speakers is still the same as before. Pretty ironic, huh. I knew my layout was iffy, but I put the regulator board in the only place I could fit in.

The photo below shows what I have. The bottom left has the filament transformer, I tried covering the leads with copper tape (that's the brown/orange stuff you see) but it didn't make any difference. The regulator board had to be put between the 2A3 and the 6SL7 sockets. So, it's either radiating stuff that's being picked up by the input stage, or it's the fact that my AC filament wires are sitting right next to the OPT/speaker terminals.

It's starting to get pretty crowded in there. And even with the heatsinks, these are getting hot. Especially the brown tie-wrapped sense resistor, I'm not sure if the plastic tie-wraps are a good idea.

Anyway... I'm going to bed now, and hopefully I won't feel so discouraged tomorrow :) Any troubleshooting advice would be most appreciated.
 

Attachments

  • curreg.jpg
    curreg.jpg
    45.5 KB · Views: 1,172
I can't solve your hum problem, but I just wanted to say that I think it is in the rca 211 data sheet we read that the ov of a dc heater supply is grounded. This results in a bias voltage of -5v in the case of the 211 (1/ the 10 volt plus whatever other bias we adopt.

In dc heating some retatain the hum dinger seen on most ac filaments. Using this would mean the bias voltage is unaltered. This is a refinement of the two resistor method.

I use the two resistor method for ac heating of indirectly heated grounds, but for my dht's I ground the 0v as per RCA. I have no idea if there is any efficacy in this, just follwing rca like a sheep!
 
I have no hum on my SE 6B4G amp. However I did, before I used star grounding, the best hum reduction method ever...

Basically, get some single core copper wire, and an earth busbar, and run one wire from the busbar to all power supply grounds, another wire to all cathode bypass grounds, another wire to all signal grounds (ie input sockets and volume pot) another to output transformer secondary, and another to chassis, and another to AC mains ground. And so on....

If you still have hum after doing that, then you may need to play around with transformer and choke placement...

I built a single ended amp using ECL82, that was much, much more crowded than your amp, and it still had no hum. It must be possible! Check your solder joints, and make sure your preamp tubes are referenced to ground somehow...
 
I'll go over my solder joints again. I also realized last night that the LT1085 datasheet said I should put a 10uF cap after the regulator, I haven't put that in there. Maybe that could make a difference. I also have about 2mV of AC ripple in my filament supply, and the cathode bias resistor is at one end of the filament (I think the high end) - could that be causing it? Grounding - that hasn't changed from what it was before. I had 2mV hum with AC heaters, and I should be able to bring that down with DC heaters, darn it :)

Thanks,
Saurav
 
OK, things are looking a little better this morning (I'm late for work because I was experimenting with the amp as soon as I woke up :)). I might have been chasing the hum in the wrong place. I put in the "Tubecad hum cancellation cap" (for want of a better term) between B+ and the cathode, and the hum went down noticably. It shows 1mV on my DMM now, which is probably the limit of its resolution, so I'll check with the scope tonight.

I had tried this cap earlier but hadn't heard much of a difference, this time the difference was quite noticable. So, I guess my filament hum is lower than it used to be, which is why I can hear the effect of changes in the B+ hum. Maybe I should try those IXYS regulators in the 2A3 B+. More hot stuff under the chassis...

... which brings me to power/heat dissipation questions. How hot is too hot on a heatsink? It's probably not clear from the photo, but I used smallish heatsinks on the Schottkys, and a larger heatsink on the regulator. I've run the amp for about 5-minute stretches so far, and the Schottkys' heatsinks get a little warm, and the regulator heatsink gets hot enough to be uncomfortable to touch. I've used mica insulators and thermal paste on all heatsinks. I think a bigger issue is airflow, some of the heatsinks are right up under the chassis.

Also, the 1 ohm sense resistor for the regulator is getting pretty hot, this is one of those cylindrical ceramic types. I have it tie wrapped to the chassis for now. Maybe I should look for metal-clad 1 ohm resistors and try and bolt them to the chassis.
 
Hi Suarav,

I had the same troubles once upon a time when I made my first DHT amp. Too much hum.

Here is what my situation was: I was using lots of filter for B+ rectification (330 µF after 1.5H choke). I was using AC filament. I was using a hum pot for the DHT filament, with cathode resistor/capacitor at the center, but at best I got about 70-100mV hum.

My solution to the problem was to use DC at the filament (bridge rectifier and 10,000µF cap) but still use the hum pot. It still had some hum at either end of the hum pot, but when set properly hum was totally gone. I mean, noise (hiss) is now louder than hum.

Why? I had been thinking about this since working on another amp with AC on the filaments and virtually no hum.

I use lots of filtering to reduce ripple on DC to almost nothing. I mean, the B+ has about 0.5mV of ripple. Therein I believe lies the problem. I tried to do this with the other DHT amp with AC filaments and actually got more hum! So, my thought is that the hum from B+ is actually being cancelled out in the output transformer by a little hum introduced at the filaments! Hence the need for the hum pot.

Since I have virtually no hum at B+, I get hum from the filaments even with a hum pot, because there is no hum opposite in polarity at the output transformer to cancel out! So my amp needs DC at the filament, with just a little bit of hum pot adjustment to completely cancel out the remaining B+ hum.

Wild and fun stuff!:nod:

Gabe.
 
I'll have to add pots to try this, because I re-used the hum pots to set the current accurately for the regulator. I also have the cathode bias resistor connected to the higher voltage end of the filament, maybe connecting it to the lower voltage end might make a difference? Or I should go buy another pair of pots.

My solution to the problem was to use DC at the filament (bridge rectifier and 10,000µF cap) but still use the hum pot. It still had some hum at either end of the hum pot, but when set properly hum was totally gone.

That's pretty much my situation now, it's equivalent to being at one end of the humpot. So maybe that's what I need to do, is buy another pair of pots.

And I have another question for anyone who could help (as always :)). I have the regulator set up as a current regulator. However, when I adjust it, should I set it for the same current in both tubes, or the same voltage drop across the filament in both tubes? I have two 2A3s, and presumably their filaments will have different resistance values when heated up. So, I could either balance the two channels by setting them both to have 2.5A through them regardless of their resistance (in which case they'll end up with slightly different voltages across the filaments), or I could set them for the same voltage drop, which will give me slightly different currents. Which would be the better solution? I think the latter, because it's current and not voltage that drives thermionic emission, right? That's harder to measure though, it's easier to measure voltage than current when exact resistance values are unknown.
 
diyAudio Senior Member
Joined 2002
Hi,

I also have the cathode bias resistor connected to the higher voltage end of the filament, maybe connecting it to the lower voltage end might make a difference?

I do recommend reading up before jumping in on the deep end, Saurav...

No offense to you personnally at all, just go through the steps like anyone does...

You really have all the answers in you're own post...

Cheers, ;)
 
I do recommend reading up before jumping in on the deep end, Saurav...

I know, I certainly didn't take my time with this one, did I :) Oh well, a couple of setbacks here and there are probably good for me, especially since I didn't blow anything up or hurt myself. I used the higher voltage end only because it was easier to reach with the component leads.

Anyway... just went and bought a couple of pots, I'll be playing with them tonight. And reversing leads to try the cathode bias resistor hooked up to the 0V line (maybe that'll work and I won't need the pots).

The rest of the heater supply circuit should stay the same as with AC, right. For instance, the only ground reference should be through the cathode bias resistor and cap, and the CT of the filament transformer secondary should stay unconnected as before. At least, that's how I think it should be, don't see the sense in having multiple ground references in the same current loop.

It did sound different though, even with the hum being the same as before, almost like there was more music present. This is a *very* preliminary impression though, because I listened for 5 minutes, and to only one speaker.
 
OK, I put the pots in, as well as the hum cancelling caps from B+ to cathode. The hum is much lower now, though not quite at the level where it's quieter than the white noise. I suspect most of the remaining hum is from the B+ (only have a CLC shared across both 2A3's B+). The humbucking pot on the 2A3 cathode doesn't seem to do much, audibly or on the scope. The hum is low enough that my scope's having trouble triggering, so I can't tell much on it anyway. My DMM says 1mV, and my scope shows a complex waveform with peaks of around 2-3mV. Anyway, I adjusted the pots for what seemed the quietest, and that's the best it can get. I think I'll need to bring my B+ hum down before I'll see/hear the differences due to the pot (i.e., moving the cathode bias resistor from one end of the filament to the other).

Before I try anything else, I need to make sure all that heat won't be causing any problems. I think what I'll do is drill some ventilation holes above the components that are getting the hottest (regulator ICs, sense resistors). Whisper-quiet CPU fans anyone? :)

Now to enjoy this for a week or two, and then go on to the next step: CCS'd ECC99 drivers.
 
Alright, I've been listening to this for 3 days now, and it's sounding really good. I also changed the cathode bias resistor from 880 to 680 ohms (I know, one change at a time, but I was in that portion of the circuit). The sound is much bigger than it used to be, almost startlingly so.

I have a question (or two). I've been playing some records that I haven't played in a long time (just because this sounds so nice now), so this may be just distortion on the records. However, I think the sound has taken on a slightly spitty, almost sibilant quality. And it feels like this comes and goes, I get short bursts of this when certain frequencies are played. Maybe I'm just imagining the whole thing. The one time I thought that it 'really happened', it sounds similar to some of the problems I had when I was building my phono stage and re-cabling my tonearm. Those I eventually traced to RFI, and solved with combinations of shielding and capacitors.

So... could this be RF problems of some sort too? I used discrete Schottkys for the DC filament supplies, so those shouldn't be generating switching noise, right. I haven't yet put in the 10uF cap after the regulator that the datasheet calls for (said it's needed for stability). Is it possible that that's what's causing it? And of course, maybe it was just distortion in one channel on that LP.

Also, I have an electrolytic cap from B+ to the cathode, and this does strange things to the output waveform. Without it, the output is almost a sine wave, kinda a pointly looking sine wave (but not all the way to triangular). With the cap in place, the output becomes a strange jagged waveform, it looks like multiple triangular waves of different frequencies and amplitudes superimposed on each other. I'm not sure what to make of that. The voltage across the cap shouldn't be going negative, because the AC B+ ripple is on some 320V DC. The cap does what it's supposed to do, cut the hum/noise level by a lot. But, I'm just curious about what that waveform means.

I also need to put the CMCs in (totally forgot I'd bought them when I was building the regulator board), and a small cap across the filament pins as suggested by someone in another thread.

Thanks,
Saurav
 
Did you use a value of 1/3 Ck for the bypass cap?

Yup. 100uF for Ck, 30 (or 33, don't remember what I found) uF for the hum-cancelling cap. It's a lower quality electrolyte than Ck though (picked it up at a surplus store, Ck is a Nichicon Muse), so maybe that's why it's doing this. I'm pretty sure I got the polarity right, or the cap would probably have exploded by now with a negative 250V across it.
 

PRR

Member
Joined 2003
Paid Member
> I also changed the cathode bias resistor from 880 to 680 ohms... I think the sound has taken on a slightly spitty, almost sibilant quality.

Get a real VTVM (or FET-VM) with a 10-Meg input and an honest-to-gosh needle.

Put it on "-VDC" and hang it on your resistance-coupled power tube grid.

Play stuff.

The needle will (better be) stuck on Zero for small signals. Big low-gain leaky tubes may show several tenths Volts DC, but it should be stable (after warm-up), at idle or at low levels.

Now crank it up.

If you get the spitty sound at the same time that the meter jumps away from Zero, you are driving the grid into the Positive Grid range. At this point the grid circuit impedance on peaks drops from 100K to 2K or less very suddenly. The driver stage just goes splat.

As for the 2A3 being hard to drive: it is and it isn't. It was made that way.

In a power triode, the lower the Mu the more peak current and power output, so you want a low-Mu tube. But the lower the Mu, the more drive voltage you need to reach full power.

With resonable assumptions, like driver B+ is lightly dropped from output stage B+, the optimum Mu is about 5.

The 2A3 runs a Mu closer to 4, because they were "always" used with self-bias which wastes a large part of the output stage B+ voltage, and the driver actually has a higher available swing.

So the 2A3 was designed to be "drivable" with about any simple ordinary triode working from the same power supply as the 2A3. Driver plate idle voltage is somewhat critical: if it is not nearly at the Maximum Swing point then it will not be able to drive a 2A3 to full output. Setting driver plate voltage to about 65% of B+ will put you in the ballpark.

Any tube has grid leakage, which sets the maximum usable grid resistance, which is pretty-much the load on the driver. The 2A3 with self-bias is allowed a grid resistor about 10 times larger than the plate resistance of an ordinary low-Mu or some medium-Mu triodes, so that they may be coupled for nearly full gain and swing.

If you use one of these new-fangled totem-pole circuits as your driver, you have a problem. The peak-to-peak swing is limited by tube drop, but now you have tube-drop on both sides of the swing (with simple resistance loading you do not have a tube-drop on the positive swing). For high-Z loads (like 2A3 and grid resistor), the total swing from a totem-pole topology is always less than a simple resistance-coupled amplifier stage. They have other advantages, and some will hold-up their large voltage swing into quite low impedances. But for driving a small "high"-power triode like 2A3 or 300B, using a totem-pole will usually demand running a higher B+ supply on the driver than on the output stage.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.