F5 power amplifier

I dont know, how much VA it is good for, but I guess more than 500VA. ....................
The inrush current may be a problem that must be carefully handled to not blow the apartment's 10A fuses
if you need ~500VA and you are on 220Vac then you should be able to close rate the mains fuse to ~T2A or at worst T2.5A. Allowing 10A from the breaker as your safety limit is nonsense.
 
I think, this 1kVA tranny is oversized.

As an alternative, I have two similar 2 x 9V 6A (108VA) trannies available.

Maybe I go with this two trannies in a double mono configuration.

While I am writing this, the postman delivered the needed prints and parts:

An externally hosted image should be here but it was not working when we last tested it.


Thanks for the comments!

Should someone (nearby) want the 1kVA tranny for free: just visit me :D

Franz
 
Member
Joined 2003
Paid Member
F-5 bias and offset

Just wondering how stable this amp should be, how much use is necessary before it gets stable, and the best way to measure offset and bias.

Upon first fireup, it will show from 17 to 25 mv, which quickly begins dropping.
Many times it will, after an hour or so, still be as high as 15 mv on the left channel, which seems to drift a little more than the right one. Last night after quite a session, (2-3 hours) one channel, right after the music stopped was around 20 mv, but came back down to around 15 mv in a few mins. I set it back to 0mv, with in a few mins it rose to 5 mv, then 8 mv, adjusted back down to 0, begins to creep back up, at 10mv I reset to 0, then it stayed around 5-7 mv.

The music playing makes it jump around quite a bit, I assume this is normal, I've noted that bias moving if music starts on my tube stuff.

Would appreciate any insight on how others amps behave in terms of offset and bias drift.

This morning I turned it on, and its sat for an hour and a half, results are:

Right channel= 13 mv Left channel= 19 mv.

What makes my bias continue to drift? Others tell me after 2 months of playing the off set remains around 5 mv.

Russellc
 
Hi,
noting the polarity of the output offset will reveal more about the behaviour with variations of use and ambient temperature.

I think this amplifier would gain much from using a good and well implemented DC correction servo/s.

It would also benefit from an alarm that activates on detecting excess output offset and, if severe, automatically shuts down the mains power input.
 
Moderator
Joined 2002
Paid Member
Offset an bias also depends on the location of the thermistors. The amp behaves differently depending on where those specific components are. If you are using a multimeter to measure offset, it should be only done at idle and with a shorted input. while playing music you are monitoring output voltage, not the offset :)

The way the thermistors work is reducing the bias level to its associated output transistor as the temperature rises. The sensitivity of the change depends on how the thermal connection to the output transistor is made - the less the coupling, the slower the change, but also longer the amp takes to warm up. There is also a bit of 'hysteresis' when adjusting with the thermistors in circuit - once the level is set, it changes a bit as the transistor warms up then you may have to back it off a bit, then up again as it may cool below the set level. Of course the offset changes with the bias levels of the two devices relative to each other, and the issue is a little more complicated by that/

I never managed to really get a result that I liked (less than +/-5mV) with the thermistors, so I removed them and now the amp does take an hour or a little more to reach operating temperature in a 25degree ambient, but the offset is rock-stable, and less than 2mV output offset with shorted input.
 
the thermistors are there to temperature compensate the bias current.
They are not there to compensate output offset.

These two requirements are quite different.
Changes in ambient temperature will change the bias. As ambient rises the devices will run hotter. That is where the Thermistors come in and attenuate the rise in device temperature with rise in ambient temperature.

The difference between the upper and lower bias currents determines the offset current and this in turn the output offset voltage.
Adding thermistors does not automatically attenuate the difference in bias currents particularly due to the delays that must be built into the feedback system.
A DC servo or combination of servos that is specifically designed to measure and correct output offset is what should be used to control offset variations, not thermistors, which as described above are for a completely different purpose.
 
Member
Joined 2003
Paid Member
the thermistors are there to temperature compensate the bias current.
They are not there to compensate output offset.

These two requirements are quite different.
Changes in ambient temperature will change the bias. As ambient rises the devices will run hotter. That is where the Thermistors come in and attenuate the rise in device temperature with rise in ambient temperature.

The difference between the upper and lower bias currents determines the offset current and this in turn the output offset voltage.
Adding thermistors does not automatically attenuate the difference in bias currents particularly due to the delays that must be built into the feedback system.
A DC servo or combination of servos that is specifically designed to measure and correct output offset is what should be used to control offset variations, not thermistors, which as described above are for a completely different purpose.

I see what you are saying and appreciate the input. As to the thermistors compensating bias and not offset, I seem to find you cant adjust one without altering the other. The bias does seem to be fairly stable. sometimes just a touch high, i.e. .61 or so. I guess maybe its not a problem.
then?

Russellc
 
Member
Joined 2003
Paid Member
Offset an bias also depends on the location of the thermistors. The amp behaves differently depending on where those specific components are. If you are using a multimeter to measure offset, it should be only done at idle and with a shorted input. while playing music you are monitoring output voltage, not the offset :)

The way the thermistors work is reducing the bias level to its associated output transistor as the temperature rises. The sensitivity of the change depends on how the thermal connection to the output transistor is made - the less the coupling, the slower the change, but also longer the amp takes to warm up. There is also a bit of 'hysteresis' when adjusting with the thermistors in circuit - once the level is set, it changes a bit as the transistor warms up then you may have to back it off a bit, then up again as it may cool below the set level. Of course the offset changes with the bias levels of the two devices relative to each other, and the issue is a little more complicated by that/

I never managed to really get a result that I liked (less than +/-5mV) with the thermistors, so I removed them and now the amp does take an hour or a little more to reach operating temperature in a 25degree ambient, but the offset is rock-stable, and less than 2mV output offset with shorted input.

Thankyou for the response. I did not know that removal of them would cause more stabilization. Does the bias (once warmed up) also remain rock solid? If I remove them, do I need to remove any other pieces, or just the thermistors? as to location, they are sitting right on top of the devices, touching them. I thought closer contact would give more stable results, but it appears my thinking isnt correct on this point. Thanks for the lesson. I dont mind waiting for warmup, if all I have to do is clip them, I think I will try to remove them.

russellc
 
Member
Joined 2003
Paid Member
in case that you want to clip them - first rebias at ,say, 30% of programmed bias .
then clip them and rebias again .

anyway - I think that entire issue about 20mV-0mV offset - no NTC-yes NTC is splitting a hair , mostly from ppl who aren't aware why NP put them in circuit , in first place .

Thank you Zen Mod! Definately a good idea to start at a low point, then adjust higher to .59...

russellc
 
Member
Joined 2003
Paid Member
in case that you want to clip them - first rebias at ,say, 30% of programmed bias .
then clip them and rebias again .

anyway - I think that entire issue about 20mV-0mV offset - no NTC-yes NTC is splitting a hair , mostly from ppl who aren't aware why NP put them in circuit , in first place .

Say, are you saying either use or dont use thermistors, or use them and dont worry about 20mV-0mV offset?

Thanks again for the response, much appreciated!

Russellc
 
Moderator
Joined 2002
Paid Member
the thermistors are there to temperature compensate the bias current.
They are not there to compensate output offset.

In my limited understanding, relative bias between the two halves of the circuit determine the output offset in this circuit, therefore when the bias drifts - and differently in the positive and negative half - the output offset also moves with it. At least, that's what my measurements showed me while I was adjusting the amp, I would welcome any corrections to the supposition :)

I did not know that removal of them would cause more stabilization.

No, not really, I have posted an answer to your other query on Peter's thread - the amp needs to be readjusted for large changes in ambient temperatures if you remove the thermistors, but (at least in my case) it drifts a little less if the ambient temperature is reasonably consistent. You can remove them and try to see if it helps with the drift - I agree with Uriah that anything below about 30mV is not a problem at all. I'm not comfortable though, as my source is DC-coupled to the F5 so I need the offset as low as humanly possible.
 
Last edited:
relative bias between the two halves of the circuit determine the output offset in this circuit, therefore when the bias drifts - and differently in the positive and negative half - the output offset also moves with it.
your first statement is correct. well at least I agree with you.
Your second statement gives the impression that output offset is somehow proportional to bias currents. I do not believe this to be the case.

The difference between the upper and lower bias currents is almost random in nature. A draught blowing more on one side of the heatsink than the other could change the bias current quite differently in the upper and lower halves. This is random not proportional.

I go back to my original two statements;
the thermistors are there to temperature compensate the bias current.
They are not there to compensate output offset.
 
To clear things a bit:
the last letter in acronym MOSFET stands for transistor - it means trans-resistor i.e. the device capable of changing its' internal resistance in accord with current running through it. So, the current through two output MOSFETS in F5 is controlled by MOSFETs' respective Vgs (and that's what NTCs and trim-pots are for).
We need the MOSFETs to have equal internal resistance (and the same current running through them) to achieve the balance which gives us zero DC offset on amp's output. Since those MOSFETs are not identical, they'll need different values of Vgs to achieve that balance. That's what trim-pots do. NTCs' job is to keep that balance regardless of temperature (since MOSFETs have positive tempco).
So, NTCs provide heat/current feedback mechanism - current running through MOSFETs generates heat and the heat lowers the resistance of NTCs which lowers the Vgs of the MOSFETs which in turn, lowers the current through MOSFETs. When precisely balanced with trim-pots and circuit is properly built, this mechanism is very accurate and not random at all. The F5 I built had the offset in +/- 2mV range regardless of ambiental influences and heatsink temperature.
 
Last edited: