I am trying to diagnose a problem with an amplifier. I was using a freq. sweep input and noticed than when I turned the gain up the I was hearing the sound. There is no speakers hooked up and the sound seems to be an exact replica. I did not try other sounds as it seems pretty obvious that something is making reproducing this sound and it is not coincidence.
I guess I could make some type of acoustical amplifier to probe around the circuit to find where it is coming from but besides that is anything I should look for? Maybe this is a natural thing? I suppose there is the possibility of a resistor or cap producing the sound due to some strange effect or it could be the sum of all the minute effects from each component but most likely it is something pretty straight forward(bad tube)?
Yeah, I competely forgot about that until I saw the smoke. Lol, luckily I seemed to have caught it before any real damage was done. Not sure where the sound was coming from but it was probably that resistor that started to smoke.
I've been using cheap medical stethoscopes for a few decades to locate the source of a sound, in electronics and while tinkering with my car. I remove the diaphragm-bell assembly and point the open end of the pipe at various parts until I locate the culprit.
Well, if you have a microphonic tube (not unusual, it's why preamplifiers often were on a different chassis than power stages), that could be getting some acoustic input and 'singing'. Is the noise broadband, or peaked at some frequencies?
Inductors can be noisy, if there's DC in them, too.
transformer plate vibrations are the most likely source. the rapidly changing magnetic field in a transformer can cause the plates to vibrate due to differences in magnetic flux, and the rapid succession of magnetic attraction and 'no attraction' (and other factors).
others have suggested microphonic tubes causing oscillation. I would actually expect that a single-ended relaxation or negative-resistance oscillator circuit might be created by open-ended amplifiers that are improperly loaded. Use a dummy load or speakers, as others also suggested. You can soak an 1 or 2 watt 8 ohm resistor (with wires attached) in a glass of water to keep it cool if you don't have a 50W resistor handy.
yes. The 'transformer buzz' you sometimes hear from power line transformers is the same kind of thing. Usually transformers are potted to limit this, since it's a loss factor also. If it's happening in a toroidal transformer, it could actually be the wires themselves (and that would be BAD because the wire insulation would eventually wear off). In any case, potting the transformer helps minimize wire vibrations also.
capacitors... maybe. Ceramic capacitors COULD become 'piezeoelectric transducers' if high enough current is running through them. I doubt this is the case in an audio amplifier. Most of the power is going through the output transformer. that's where the sound is coming from.
Is it singing the words or just humming? If it's humming it probably doesn't know the words. In that case you should start singing a few bars and give it a chance to learn the words. I think then it will be OK.
The output transformer has a number of primary and secondary windings and they always make some audible sound even with no speaker connected. The reasons are that the magnetic currents cause microscopic movements of the coils and core structure an the voltages between anode wndings with high audio frequency voltages. These Vac act into the capacitances and you get an effect like that in an electrostatic speaker. So the OPT acts like a transducer in more ways than one.
When you have no speaker connected and no dummy load the OPT signal voltages can me made to rise up to about 3 times the maximum output signals when a load is connected. This has been said to occur because of back emfs involving the leakage inductance, or merely the back emf with each half primary when the idle current cuts off. When one of the 2 tubes cuts off its current when the grid goes to a low negative value, the anode voltage rises because that's how inductors behave. ( if you don't understand this very simple basic property of inductors+current change then don't be lazy, do some study via Google or some books ) The other tube is being turned on with a positive going grid voltage but its anode voltage moves towards 0V and continues below 0V to maybe a negative voltage =3D twice the supply voltage, and during this time this tube is cut off, even with a positive going grid. The absense of a load on the OPT allows this huge Vswing of perhaps up to 3 times the normal Vaa with a load connected. Normal Vaa might be 500Vaa rms, and you might measure 1,500Vrms without a load before anything clips. Such a huge Vaa can cause arcing across poorly insulated windings or at output tube sockets, typically between pin 3 anode connection to pin 2 filament connection which is earthy.
Perhaps you may be unaware of the high voltages you might be producing when you have no speaker connected.
Beware, you might damage your test gear, and damage your OPT!
There is a solution to the problem of excessive anode voltage generation when no load is used. Connect two simple networks of 3 x 1n4007 diodes in series all facing in the same current direction. Connect the end diode anode of each diode network to the output tube anode. Connect the end cathode of each diode network to 0V. Normally the diodes never conduct any current. But when one side of anode winding tries to move below 0V because of the back emf in the other half primary, the diodes conduct and harmlessly shunt the low value of current causing the back emf. So the Vaa swing at each anode is therefore limited to +/- the peak voltage equal to the B+ at the CT. This action is due to what is called a diode clamp, and "clamping diodes" should be fitted to every PP tube amp to prevent excessive Vaa occuring. IN larger amps in ther past when B+ may have been 2,500V with larger transmitting tubes, silicon diodes would be useless. One common way to prevent excessive Vaa was to have a pair of adjustable sparking points connected anode to anode with a load resistance of say 10k, so that if Vaa went over say
3,000V, then there could be an arc between the points and the current flows through the 10k load and damps the Vaa. The distance across the gap is adjusted just right.
Its natural, and you can't avoid what is called "OPT howl". If anything, SE trannies howl more than PP trannies. Potting the OPT reduces audible howl which can be quite loud while playing music.
Excessive howl is produced in poorly varnished transformer windings and poorly clamped up laminations.
Tubes, caps and resistors make a tiny amount of howl. 95% of the howl is due to the OPT.
On 01/15/11 15:54, Patrick Turner so witilly quipped:
oops - I think I understand what you're trying to do here and the diodes should be the other way 'round. So if plate voltage goes below ground potential, you want the diodes to conduct, right? That way you're preventing overly negative voltages on the transformer primary.
But if you have a negative F/B loop driven from the secondary, I think that would solve the problem too, as in no 'infinite gain' on the power tube stage. You may still have a problem with spurious oscillation due to negative resistances and Z mismatch.
FYI your explanation as piezioelectric rather than magnetic vibration is interesting. I would have expected plate vibration to be the primary source, and counter EMF on the windings (like you get in a motor or generator) to be the secondary factor (for wire motion). Still, no matter what the root cause (which appears to be a number of things, no surprise here) potting is the solution to prevent it and reduce losses.
Its easy to become cornfused in the aftermath of the silly season.
If anyone did what I said the diodes would short the B+ supply to 0V at each end of the OPT.
One could be a purist and use vacuum diodes as the clampers, but you need biased heater supplies, and single diodes, and its all too much trouble, so nobody ever did it. Makers just instructed owners to always have a speaker connected. The other thing that often happens without a load in an amp with NFB is the problem of LF and HF oscilations because the open loop gain rises without a load but the amount of effective FB increases. Many amps made by ppl who don't understandunconditional stability will oscillate with no load.
I think one will find the NFB does not cure the problem and the secondary voltage will rise enormously above normal levels with no load especially with a pure tetrode/pentode configured amp. UL and triode connection reduces the problem though.
Try testing a tube with a resistance load only to a high value B+. You should find the anode hardly makes any noise at all. A choke makes very little noise because there is little capacitive interface between high voltage swings and something earthy. But in a OPT there are highly changing voltages against earthy secondaries, and things try to move a little, like in an ESL. But the majority of howl is magnetically caused.