Welcome to Solid State Guitar Amp Forum | DIY Guitar Amplifiers. Please login or sign up.

March 29, 2024, 12:47:13 AM

Login with username, password and session length

Recent Posts

 

hartke3500 blown transistor

Started by js1970, October 12, 2014, 08:00:21 PM

Previous topic - Next topic

Roly

Quote from: js1970I'm reluctant to go full mains.
    Why does the bias drop like that with increased signal?
    You said the bias is a static setting, which i take to mean it is set  for the power amp to run most efficiently to amplify a signal, but with the signal effectively stealing voltage from this setting, how can the power amp be working sufficiently, let alone efficiently?

Uh huh, this is about what I'd expect.

Limiting lamp, increased signal, increased current, supply voltage droops, bias circuit responds.

Quote from: js1970I ran the hartke through a 100W floodlight bulb

In my workshop if that didn't produce smoke I'd call it "fixed".   :dbtu:




To recap; the bias is required because the transistors in the top and bottom of the "totem poll";



... don't start to turn on at zero volts but about 0.6 volts, making a deadband of around 1.2 volts while the signal crosses zero.



This is extreme Class-B and produces horrible crossover distortion.



To eliminate this the output bases are driven with an inbuilt bias between them to bring both to the edge of conduction while idle, and to ensure that there is a smooth change over of conduction between them at signal zero crossings.



Away from zero the bias has little effect since one or other of the output transistors will be in major conduction and the other off, even 'tho "biassed".


{if you really want to get into the nitty-gritty you need to look inside the devices, at their models. 



Particularly with power devices, you don't just have an "ideal" transistor.  In this case you need to consider that each electrode has its own parasitic resistance in series, so that when the base current goes up the actual pin voltage will be somewhat higher than the theoretical B-E voltage of the ideal (internal) device.  But you can't actually get to that, you have to deal with the real-world at the leads, and that includes electrode bulk resistance that gets significant under some conditions.

Similarly a power FET may look like an ideal switch under DC conditions, but the gate has a huge parasitic capacitance, so if you want to run one in Class-D at 100kHz clock rate you have one hell of a problem driving that gate capacitance fast enough so it doesn't get horribly inefficient and lossy.

Water is wet, rocks are hard, and devices have real-world limitations you just have to work around.}

HTH
If you say theory and practice don't agree you haven't applied enough theory.

js1970

  Thanks Roly,for the info. Your patience hasn't gone unnoticed or unappreciated.
  With that being said....
  I ran the amp at full mains ,only to have it overheat. After about 15 min, the protection relay clicked open, and the heatsink was hot to touch.
  I let it sit awhile and checked everything through the limiter.
   TP1 to G ~ 8.6 mV
   TP2 to G ~ 6.2 mV
   TP1 to TP2 ~ 2.6 mV
   Halfrail 6.6 mV
  I think these are good, yes?
  Then I disconnected the limiter and ran full, btw TPs it started at roughly 63mV and was rapidly rising. I shut er down without the other supporting measurements.
  So basically, the amp tests good under limited power, and drives hard on full.
As I'm writing this, l can't help but wonder if maybe R321 or R317 has gone bad. My thinking being, all values are good with the limiter providing the resistance needed. Take away the limiter....danger Will Robinson.
   

 

Roly

Quote from: js1970l can't help but wonder if maybe R321 or R317 has gone bad

Well don't just stand there muttering "short circuit" lad - lengthen it!  Out with your multimeter and prove the point to your satisfaction one way or the other.


Quote from: js1970I ran the amp at full mains ,only to have it overheat. After about 15 min, the protection relay clicked open, and the heatsink was hot to touch.

I assume this was at or near idle, not fully driven.  If 15 mins of full drive opens the thermal cutout then I wouldn't be impressed, but I also wouldn't be too worried (and I'd fit a fan).

But the thermal tripping after 15 mins only on idle is a real worry.

It could be that the idle bias is set too high, but it certainly seems to be that the thermal compensation, thermistor TH302, which is supposed to reduce the bias as the rig gets hot, isn't doing it job.

This thermal compensation is more normally a couple of forward-biased diodes.  As they get hot the idea is that their forward voltage drop falls in step with the B-E junctions of the drivers and OP transistors, and ideally the bias current remains rock steady from freezing to roasting.  Ideally.

In more professional rigs you see these compensation components actually mounted on the heatsink, which makes very good sense since that is what they are supposed to be responding to/measuring.

In consumer stereos they are almost always fitted on the PCB nearby on the assumption that, given the amplifier duty, they will be about the same temperature.  In high power rigs this is not a safe assumption, and simply having it out of its case can effectively decouple the sensor from the thing it is supposed to be sensing.

Now you previously observed that heating the thermistor with your fingers caused the bias current to drop, so that suggests that the circuit itself is working properly, just that it isn't having enough effect to hold the bias under control.


However;
Quote from: js1970the protection relay clicked

All this does is look for standing DC on the output, it isn't the thermal cutout.

So I'm wondering; you previously set the bias current, but wasn't that with a Limiting Lamp in series?

Perhaps you should back the bias off, connect the amp direct to mains, then carefully re-set the bias to the correct value on full mains voltage, and closely observe it as the rig warms up.

Let's see where that leaves us.
If you say theory and practice don't agree you haven't applied enough theory.

js1970

   Heh heh, yes I hear ya. I metered all of the 2W resistors to find good values. R317 ,however, was super hot and was found with 83V and 43V on the legs relative to ground. This was at idle under full power and bias set.
   I reset the bias like you said. At idle btw the bases of Q309/310 I have 2.540V and trending down as it warmed up. Also, btw the TPs, I started at 24mV which dropped steadily to settle about 16 mV +/- 1mV.
   These measurements were taken over 30 min at idle. The components were slightly warm, except for R317 which was very hot.   
   Should I be concerned that this resistor got hot,or that the others did not?
   Does it seem like a lot of voltage on it?

Roly

Quote from: js1970I reset the bias like you said. At idle btw the bases of Q309/310 I have 2.540V and trending down as it warmed up. Also, btw the TPs, I started at 24mV which dropped steadily to settle about 16 mV +/- 1mV.
   These measurements were taken over 30 min at idle.

This all looks quite good indeed.  The bias may be a bit over-compensated for temperature, but that's not particularly abnormal and far better that than going into thermal runaway.


R317
What we've got here is a differential or Long Tailed Pair, Q304 and Q305 (common tail above, loads below).  The Q305 side is the drive for the power amp, the R317 side its differential partner.  Now what they have done is run these two LTP loads into a current mirror, Q306 and Q307, and if you suspect that something is amiss I would have these transistors out for checking (however it is not uncommon for power resistors to run too hot to touch.)

{Note that the preceding stage is also a LTP but the other way up, and a current mirror has been used to set the tail current here.}

To be specific: on the circuit I'm looking at <Ha3500.pdf> R317 is a 15k/2W in the collector of Q304.

83v and 43V on 15k

I = E/R
(83-43)/15 = 2.66666667mA

P = E * I

(83-43) * 2.66666667 = 106.66666680mW

This seems very reasonable, a tenth of a watt ain't much, so something doesn't fit here, your observation doesn't make sense with my circuit.
If you say theory and practice don't agree you haven't applied enough theory.

js1970

   I did some research and looked into the Long Tailed Pair circuit. I'm trying to get the gist with regards to what you said. I'm not sure what doesn't fit. Is it that I'm getting 40V difference at R317 ,or that if 2.67 mA are flowing through it,it should not be so hot.
 

Roly

The Long Tailed Pair, a.k.a. a differential amplifier or emitter/cathode-coupled amp.


(the interlocking circles at IE is a constant current sink, but often just a resistor)

The "long tail" comes from the shared emitter (cathode) resistor.  This is normally a high value connected to a fairly high negative voltage, and therefore approximates an infinite resistance connected to an infinite negative voltage, the result of which is (almost) constant current.

A refinement is to make this common tail an active current sink (source) so that within the limits of available voltage it really does act like and infinite resistance connected to an infinite negative voltage.




The "differential" comes from the fact that it amplifies the difference between the pair Bases (grids) and also has a differential output available at the collectors (anodes).  As such it is a very sensitive amplifier, 'tho easily driven into overload.  Ideally the diff. pair are quite deaf to any signals in common mode to both inputs.


Because the total tail current is constant, so therefore must the sum of the Collector currents be.  As one collector current rises from zero so must the opposite Collector current fall by exactly the same amount, so the Collector currents (and voltages) mirror each other.


The "Emitter/cathode coupling" comes from the signal passing from one device to the other via their shared Emitter/cathode circuit.

The LTP structure forms the basis for many op-amps (look at op-amp datasheet internal circuits).




In audio power amps the signal input is generally the non-inverting (+) input while the other is the inverting (-) input and is connected to the main output to provide overall Negative FeedBack.

Because of the high gain the LTP is sensitive to any difference between the two active devices, in particular the thermal change in the Base-Emitter voltages.  For this reason they are sometimes matched pairs, and you encounter twin transistors with six legs so they are tightly thermally coupled (apart from being inherently matched twins on a common chip).

{on the other hand Elektor made use of this "shortcoming" by using two fixed-bias LTP's at right angles  as a no-moving-parts wind speed and direction sensor, the wind cooling the upwind device more and producing a voltage proportional to the vector of the wind.}


Quote from: js1970I'm trying to get the gist with regards to what you said. I'm not sure what doesn't fit. Is it that I'm getting 40V difference at R317 ,or that if 2.67 mA are flowing through it,it should not be so hot.

Simply that your two observations are contradictory.

Your voltage measurements suggest that the resistor is dissipating one-tenth of a watt (and these numbers look pretty sensible), which is almost nothing and you should be able to hold the resistor between your fingers and feel a bit of warmth.

Yet the resistor is getting very hot.

These can't both be right.

I'm inclined to place more trust in a burnt finger than your measurements or my calculations, but there is an anomaly here, and "following the bit that doesn't fit" is always profitable when faultfinding.

I can't guess what the answer will be, but these two observation are uncomfortable bedfellows - there is something that needs to be resolved here.
If you say theory and practice don't agree you haven't applied enough theory.

js1970

   After reading your last post, I diode tested the transistors and they seemed up to snuff. So I cleaned off the cooling fan, fastened everything down tight, and ran it. I had to keep the amp at sensible volumes, but nevertheless, I ran it for about an hour. I must say that thing is a beast. Punchy, defined, with gobs of power. At times, it seemed almost too punchy, too powerful ( if indeed, there is such a thing).
   Satisfied, I took the lid off and checked the resistor with the back of my finger. Barely warm! Maybe my original observation was a bit askew, or maybe the thermal management of the fan and heatsink made it a non issue. I'm going to take it  to a jam and see how she does.
    Roly, I'd like to thank you and all from the forum again for your help with this fault finding mission. I like to think I learned a little bit, and fear I have set on a new hobby to which my lovely ,understanding ,bride will be thrilled.

Roly

As long as somebody is receptive I'm happy to go on 'splainin' 'til the cow come home.


Quote from: js1970After reading your last post, I diode tested the transistors and they seemed up to snuff. So I cleaned off the cooling fan, fastened everything down tight, and ran it. I had to keep the amp at sensible volumes, but nevertheless, I ran it for about an hour. I must say that thing is a beast. Punchy, defined, with gobs of power. At times, it seemed almost too punchy, too powerful ( if indeed, there is such a thing).

That's "fixed", yup.   :dbtu:


Quote from: js1970Satisfied, I took the lid off and checked the resistor with the back of my finger. Barely warm! Maybe my original observation was a bit askew, or maybe the thermal management of the fan and heatsink made it a non issue. I'm going to take it  to a jam and see how she does.

A bit of (forced) air movement over a component can make a very large difference to its temperature, but it's this final good result that is the key point.


Quote from: js1970fear I have set on a new hobby to which my lovely ,understanding ,bride will be thrilled.

Ahhhh.  Now, a couple of tips.

The first is that electronics, building or faultfinding, is actually quite addictive.  My wife (my third :o) is a puzzle buff, crosswords, Suduku's, and the like.  I get a similar charge from solving faults on the bench, and this "charge", endorphins or whatever, is quite addictive.


Secondly, electronics generally isn't horribly messy like working on car engines or building surfboards, but it can be a bit ... untidy.

{My wife's ex worked on big engines and had one in bits in every room of the house except the bedroom, where he had been clearly warned, so when she came home and found another one pulled down in the bedroom and grease on the bedsheets that was it.  Finish.}

For a time we lived in a small flat and there was nowhere that could be set aside for working on electronics projects, so I had a large cardboard box I could put on the dining table and work on stuff without any danger of damaging the surface or leaving solder splashes everywhere, and which could be packed up and hidden away to restore what girls like to think of as domestic normality.

Some flat dwellers have even built specific boxes that open into a mini-workshop or Ham station where it can all be hidden away.

So my hints are;
1. look after her in bed.
2. promptly fix everything that isn't working properly, blinds, heaters, door catches, etc, and she'll be a lot more tolerant of your electronic man-interests.
3. Chocolate.  It's like beer to us or catnip to cats, but only in small quantities of the highest quality you can find.  Here we have "Tim-Tams" which are not particularly up-market, but the dark chocolate ones are very effective at calming upset partners;



And if she ever starts complaining about "all these wires everywhere" gently remind her that you don't work on cars (and make certain that nothing electronic ever turns up in your bed - they really don't like that).

It helps a lot if she's into something creative herself such as music, photography, painting, dressmaking, pottery, writing that killer novel, radiation polymer chemistry, &c&c.

4. don't take my advice on relationships.

ATB to you both.

https://www.youtube.com/watch?v=m2ou-WIxfLY
If you say theory and practice don't agree you haven't applied enough theory.