Not a problem. As the DHT in question is operating in Class A, the interemod product amplitude arising in each section will vary only a few percent. So the intermod balancing at high level is not perfect, but it is still effective.
And with a high level signal, ear masking is even more effective than at low level.
So - finally you admit that the DHT has a mechanism for generating intermodulation, regardless of nulled hum.
All that remains is to quantify it. Any Constructor can easily do this for themselves - it only requires a sound card & installation of an FFT app.
The IMD will vary from tube to tube, and with the implementation - but will always be there with ac-heating, and always be negligible when properly implemented dc supplies are used for HV and filament heating.
I am glad we now have that out the way.
Perhaps we can now give Joel his thread back, and help him to solve his noise problem.
Well, is there something different about the directly heated tubes in an AM transmitter modulator and directly heated tubes in a audio amplifier then? Apart from size?
There isn't, of course. They work in the exact same way.
Actaully, that employer had a whole chain of AM and FM transmitters across the country. Those STC transmitters gave a good acount of themselves, and were still in use quite recently. On straight AM their sound was excellent. Unfortunately when AM stereo arrived in the 1980's modification for stereo was problematic and distortion on AM stereo was never really acceptable. They were also expensive to run, consuming about $20 worth of electricity (today's dollars) an hour (~$3000 per week) and virtually requiring full time technical attendence. What killed them off was the unavailablity of the RF power tetrodes after the 1990's.
There were always mechanisms - some others were identified by others in other threads.
But my argument that when hum is nulled out, the hum intermod will be inaudible still stands.
You can try this test yourself (remembering that the intermod products must be lower in amplitide than teh hum itself anyway) to verify ear masking.
Get two audio sinewave oscillators. Use one as the input to one channel of a stereo amp, and the other osc as the input to the other channel. Place the speakers next to each other, so your ears can't use directional cues.
Set one to 100 Hz and the gain so the the level heard in the speaker is at a comfortable level, not overloading. Set the other osc to 133 Hz and at a level say 30 dB lower. 40 dB lower if you think your ears are good. Turn this osc's output on and off repeatedly, using output level pot (so that there are not switching transients that the ear can detect) - you should be able to hear a clear tonal character change.
Repeat the experiment with 1000 Hz and 1330 Hz. You should find the tonal character change obvious and not particularly pleasant.
Repeat the test with 1000 Hz and 1030 Hz. You will be battling to hear when the second tone is switched on and off. If you do hear anything it will be due to the very small net power level change as much as anything.
It may be difficult with many low cost oscillators and function generators to set them close to 30 Hz apart. This can be resolved by temporarily setting both to about the same level, and adjusting one so that the beats are just no discernable.
Yoy can of course program a PC sound card to do this. But bear in mind that the test is invalid if the sound card is operated near clipping, and cheap sound cards have all sorts of obsure distortion, bus hash, and D/A artifacts that cna confuse the issue.
Go and try this experiment. Then come back, tell us HONESTLY what the result was.
I've demonstrated this in front of others, with both oscillators fed into a mono amplifier. None could tell when the 1030 Hz was on or off. The oscillators werer designed for speaker testing and had zero crossing switching so no transients were induced.
Last edited:
Sure. You started this nonsense, Rod.
You can give up your wild claims, never at any time substantiated with technical explanations, any time you like.
So, your technical argument has collapsed, you admit the IMD will exist (although you are unable to measure it) but now you want us to do a listening test.
OK we need to do this, but I suggest a far more relevant and useful test: simply take a high quality DHT amplifier, and compare ac-heating to properly-implemented dc heating. Even with voltage regulators (though further improvements with current-drive will be easily heard). No raw rectified dc though.
Anyone can do this, and many have reported their findings here on DIYA.
In your case, you'll first need to actually design & build a DHT amplifier, of course, rather than any more relying on your imagination.
OK we need to do this, but I suggest a far more relevant and useful test: simply take a high quality DHT amplifier, and compare ac-heating to properly-implemented dc heating. Even with voltage regulators (though further improvements with current-drive will be easily heard). No raw rectified dc though.
Anyone can do this, and many have reported their findings here on DIYA.
In your case, you'll first need to actually design & build a DHT amplifier, of course, rather than any more relying on your imagination.
Never said it didn't. I said if the hum itself is inaudiable, the hum/signal intermod will also be inaudible, and in fact will also be much lower in measured level than the hum itself. Nothing you've said refutes that.
I explained why, mathematically this must be so. The only thing you've thrown in is a bit about high signal levels disturbing the hum balance. But with high signal levels the tube's gm will be affected all along the filament, so any imbalance is only a secondary effect. Since the gm is altered only a few percent, the gm unbalance must be even less. Quite a bit less in fact.
You can verify this yourself, just do what I and others have sometimes done when a high power amp is on he bench: With the tube bias at the proper level, adjust the AC heating balance trimpot for hum null. Then shift the tube bias to whatever off-normal level you like, so long as the tube is not near cut-off or bottoming. Does this make the hum re-appear? Never more than at tiny levels - if at all.
You still haven't disproved ear masking.
Well, have you done that, then? I suspect not - you talk hot air.
As I said before, only an idiot would used unfiltered rectified heating. That will give you volts of harmonic-rich "signal" that cannot be nulled out.
The intent of the experiment I suggested is to show you ear-masking in action. If you can't hear close-in intermod products (at levels, albiet low, considerbaly exceed what actually happens), it doesn't actually matter whether they get produced or not, does it?
Are you afaid to do the ear masking test, perhaps you percieve that you are wrong. Or lack the test gear to do it with? That wouldn't surprise me - I find that the less instrumentaton people have, the less insight they have and the more likely they repeat misconceptions.
Why are you so so keen to push regulated DC filament energisation? Are you selling the regulators?
Last edited:
It's not cooperating because of a lack of experience..
I have modern x1 / x10 probes.. I don't think they're HV. When you say a well insulated wire, do you think just some heatshrink covering the end of the wire which I'll wave around the anode lead will be okay?
The wires from the HV power transformer are indeed quite long and spaced apart.. Maybe I'll have to rethink the whole layout on the board.. Unfortunately due to the size of the components it's really hard to keep lengths short.. When I put this into a chassis I'll be able to mount on the top and underside of the top plate obviously, so wire lengths will be able to be much shorter.. But I can probably figure something out to rearrange things on the board.
I'll also put the third HV cap near the cathode resistor.
The usuall x1/x10 and x10 probes sold with and for oscilloscopes are usually rated for only 400V, sometimes 300 V. Perfectly adequate for working on solid state equipment.
Several manufacturers make special probes for use on vacuum tube equipment and high powered AC motor VSD's, and these can have ratings up to several kV. They are clearly marked as such and often have a distinctive appearance. The 1.3 KV ones I have (made by Sefram in France) do anyway.
If your probes have no markings as to voltage rating, best to assume they are only good for 300 V or less.
Heatshrink over the wire should be fine.
Several manufacturers make special probes for use on vacuum tube equipment and high powered AC motor VSD's, and these can have ratings up to several kV. They are clearly marked as such and often have a distinctive appearance. The 1.3 KV ones I have (made by Sefram in France) do anyway.
If your probes have no markings as to voltage rating, best to assume they are only good for 300 V or less.
Heatshrink over the wire should be fine.
Last edited:
Don't forget that a loop around which a current flows (a transmitting loop), can only induce a noise if there is a receiving loop that the magnetic field can traverse.
You said that 1) shorting the grid resistor stops the noise. That indicates the "receiving loop" is grid-grid resistor-cathode bypass-cathode-back to grid. This is not significantly altered by shorting the resistor, unless you did it by a direct short from cathode to grid on the socket (which would of course turn the tube hard on so it couldn't amplify).
You said that 2) the noise is heard in the tweeter, which indicates it is made up of high frequency energy. Loops tend to couple low frequency.
I'm not saying Rod must be definely wrong here - I just don't want you to get your hopes up on a lot of work and then find it didn't help much. Try easier things first.
Last edited:
When I shorted I clipped from where my signal input would connect to, between the grid stopper and grid leak resistor, and then grounded that to power supply ground. Not directly to the cathode.
So a loop problem is likely to be only part of the story.
The interesting bit is where you say "...and then grounded it to the power supply ground." By a longish wire perhaps? You may have substituted the existing "receiving loop" with another one. And by shorting the stopper, you mave have helped the tube to oscillate, repalcing one buzz with another one.
What happens if you directly short the grid end of the stopper to the earth end of the grid leak resistor? With a short direct wire? Or if you just directly short the grid leak resistor and leave the stopper resitor in circuit?
What happens if you directly short the grid and filament -ve with a capacitor, say 0.1 Uf (100 nF), (the value doesn't matter much), which won't upset tube operating conditions? Keep the cap leads as short as possible. Install teh cap RIGHT ON THE SOCKET. DO NOT use clip leads. The answer could be quite revealing. Especially if you can prove the tube is not oscillating.
By now you should be getting a sense of the two approaches to problems: Rod is telling you good construction practice but flying a little blind. I'm teaching you to think, do tests, reason out the cause, and apply a specific cure. Both approaches have merits.
Last edited:
That's hilarious. For someone caught with their pants down making claims about IMD level, and yet unable to provide measurements when challenged - who are you to talk about lab gear? There's no sign that you even have a basic FFT setup.
For that matter - when have you ever posted any measurements of an amplifier YOU designed? In any of your hundreds of rambling posts? We already know you have no DHT amplifier to measure. It would be unsurprising to find you have never designed and built anything resembling real high fidelity.
Rod, it's very revealing that you can only make these sort of emotional attacks.
Not once have you provided any logical technical discussion. Not once have you ever come up with any reason why my reasoning is wrong, other than that bit about large signal imbalance, which I showed was a non-issue.
As a professional engineer, I have and always have had, a access to a full range of instrumentation owned by employers. As you would expect. And now that I am working as a consultant, I have a full range of gear I own. Including both spectrum analysers and a wave analyser. Wave analysers are instruments that offer greater sensitivity and precision in audio applications than do typical spectrum analysers.
Now, do you have the gear or don't you?
Not once have you provided any logical technical discussion. Not once have you ever come up with any reason why my reasoning is wrong, other than that bit about large signal imbalance, which I showed was a non-issue.
As a professional engineer, I have and always have had, a access to a full range of instrumentation owned by employers. As you would expect. And now that I am working as a consultant, I have a full range of gear I own. Including both spectrum analysers and a wave analyser. Wave analysers are instruments that offer greater sensitivity and precision in audio applications than do typical spectrum analysers.
Now, do you have the gear or don't you?
Last edited:
Well, in the first place your claim was extremely funny, in the circumstances.
The rest of my response was hard fact.
The bit about the large signal imbalance is the whole point. You have not shown it to be a non-issue: it's what leads to the IMD measurements we have seen.
The rest of the fluff you raised is irrelevant, and what you call logic is, as I have demonstrated, based on bogus assumptions.
No evidence of it, in any of your posts, being put to use.
If you wish to be taken seriously, back up your claims with measured evidence. That's the scientific approach - an approach which you are clearly not used to.
Oh, maybe I wasn't totally clear.. When I do ground between the stopper and leak resistor to power supply ground, the buzz is completely gone.. And I mean totally completely gone.
I'll try some of the other suggestions later this afternoon.
As for your suggestion to encase the grid resistors in a metal box and ground it.. My original statement that it didn't make a difference was wrong.. It was subtle enough to my ear that I didn't notice it.. Although when recording the difference I can tell that there is a clear difference.
Here's an audio clip. In this clip I have the grid resistors encased in the metal box, and it starts out with the box grounded, about 10 seconds through I unclip the box.. It's a noticeable difference.. Although part of me wonders if having an ungrounded metal box around the grid resistors causes some additional noise that I wouldn't have had with no metal box there.. It's pretty much impossible for me to totally remove the box while the amp is live.
https://soundcloud.com/thehoj-1/buzz4
Also, here are a few video clips of my oscilloscope display with that probe measuring some different locations.
I apologize for how difficult it is to see the display..
In this clip I have the amplitude set to 50mV/div, so fairly substantial I guess. At about 25 seconds I move the probe near the rectifier tubes, and it picks up some large looking spikes..
https://youtu.be/AtjDj5DCYFk
Here's another clip where I move the probe to the anode of the 845.. It's touching the anode pin on the socket, and the amplitude is set to 2mV/Div in this video.. There's a small waveform that shows up, but barely noticeable.
https://youtu.be/q7WSbmC2x3s
Yes, the rectifier and surrounding region will give out plenty of emissions. That scope-shot is quite normal.
This is the reason to try to keep the trafo/rectif./caps at a safe distance from the sensitive circuits - and why the loop area of all the rectifier/trafo/cap wiring must be minimised, for low noise.
This is the reason to try to keep the trafo/rectif./caps at a safe distance from the sensitive circuits - and why the loop area of all the rectifier/trafo/cap wiring must be minimised, for low noise.
Disregard my previous post. I was trying to trick you. Well, actually I had a senior's moment - for some reason I went from thinking that shorting the grid resistors made a great diffreence to thinking it may hardly any.
Correct. An unearthed box adds another problem.
Spikes like this are what you would expect near rectifiers and what you've got is not particularly significant.
This is rather indeterminate.
It looks like some RF, but it could be the tube is picking up a radio station, or it could be the tube is oscillating well up in the VHF region where the oscilloscope isn't able to respond very well. The Philips PM3234 has a bandwidth rather low - only 10 MHz as I recall.
Try earthing the LO end of the grid leak resistor, with the earth wire just as you did when you shorted the resistors, but this time just have the extra earthing without shorting. What happens to the buzz?
The evidence is building up that more than one problem is contributing to the buzz.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.