Charles,
I thought I was almost the only one here using such old
equipment.
Until a year ago I had a CDP I bought in 1986.
I think it was a quite good one back then and probably did
stand the time reasonably. It was one of the very few not
exremely priced CDPs with a true 16-bit DAC. Most were 14 bit
with oversampling, as I am sure you know. I finally replaced it
a year ago, when my best friend got a bargain on a SACD
player and I took over his not very old Sony XB920 for a
bargain. With such friends, who needs shops?
John,
are you saying that despite all the LP filtering on both digital
and analog side, the LP filtering at several 100kHz still matters?
I haven't tweaked my new CDP yet, but have the service
manual and have been thinking about what mods to do.
Of course, removing the muting transistors is one that has
come to mind. Does one ususally get much problems with
pops and clicks if removing them?
I thought I was almost the only one here using such old
equipment.
Until a year ago I had a CDP I bought in 1986.I think it was a quite good one back then and probably did
stand the time reasonably. It was one of the very few not
exremely priced CDPs with a true 16-bit DAC. Most were 14 bit
with oversampling, as I am sure you know. I finally replaced it
a year ago, when my best friend got a bargain on a SACD
player and I took over his not very old Sony XB920 for a
bargain. With such friends, who needs shops?
John,
are you saying that despite all the LP filtering on both digital
and analog side, the LP filtering at several 100kHz still matters?
I haven't tweaked my new CDP yet, but have the service
manual and have been thinking about what mods to do.
Of course, removing the muting transistors is one that has
come to mind. Does one ususally get much problems with
pops and clicks if removing them?
johnrtd said:
the voltage-controlled capacitance on a (normal) diode is very very small. In my previous life, I had built FM receivers based on the technique, but used diodes specifically designed for this purposes because normal reverse-biased diode wouldn't be large enough.
you are talking about FM signals in the 90Mhz and above.
I wouldn't be too worried about this at audio frequencies.
It seems that a thermal equilibrium comes back again and again as being responsible for the start up effects.
The thermal model of an amplifier might be drawn to determine an optimal thermal startup response.
We mainly looked at class-A/AB amplifiers. If we look at amplifiers working in a complete other way do they behave differently? Does a class-D amplifier sounds differently directly after switching on and after "warming-up"? And if so how much time does that take? Anyone having experience with this?
Ward
The thermal model of an amplifier might be drawn to determine an optimal thermal startup response.
We mainly looked at class-A/AB amplifiers. If we look at amplifiers working in a complete other way do they behave differently? Does a class-D amplifier sounds differently directly after switching on and after "warming-up"? And if so how much time does that take? Anyone having experience with this?
Ward
EDUM said:It seems that a thermal equilibrium comes back again and again as being responsible for the start up effects.
The thermal model of an amplifier might be drawn to determine an optimal thermal startup response.
We mainly looked at class-A/AB amplifiers. If we look at amplifiers working in a complete other way do they behave differently? Does a class-D amplifier sounds differently directly after switching on and after "warming-up"? And if so how much time does that take? Anyone having experience with this?
Ward
Part of a Class-D amp is analog as well!
John
Sure it is. But as the low power audio signal is treated in a different way and the ouput stage operates completly different from an class A/AB design the audible effects during start up may be different or completly not there. Further as a Class D is verry efficient the thermal behaviour, what we indicated as a possible source for all te effects, is less important.
I only want to say that with the feedback it is not a simple mechanism. You do not want a system with a long settling time and large overshoots. Certainly not with an audio signal disturbing this equilibrium
It might be worth to spend some time on it in the design process.Ward
I completely agree with that and this more or less the same as I already mentioned before, just in other words.
The thermoelectrical feedback-loop used to stabilise the bias-current is certainly less easily modeled than the electrical part of the amp.
So why in all the world bother with that at all, when it would be possible to stabilise an electrical property by a purely electronic feedback loop ?
Regards
Charles
Accurately controlling things like temperature and bias would be relatively easy to do, but I'm not sure this necessarily makes the amp better.
In order to keep one particular parameter constant we have to change another, so it's important to ensure that whatever changes is less important than what we want to keep constant.
In order to keep one particular parameter constant we have to change another, so it's important to ensure that whatever changes is less important than what we want to keep constant.
Noize's thread seems to fit with some anecdotal experience I'm having with an Adcom GFA-535. (It needed a re-cap, bias pots, etc) The manual says to warm the amp up at full power for 5 minutes at 8 ohm, then set bias across the emitter resistors to 5mV. This gets it pretty dang hot! As it cools again, the bias drops to only 1mV, and it actually showed 0V for perhaps 30 seconds after turn on from dead cold, before settling around 1mV.
So I'm a little unsure what to do here. I've let the amp warm up for a few hours now, and I can stabilize it at 5mV with no signal, and if I get it hot driving a dummy load, it goes up to 17mV. This seems a pretty reasonable method, but I have not done a lot of this kind of work. Opinions, please?
So I'm a little unsure what to do here. I've let the amp warm up for a few hours now, and I can stabilize it at 5mV with no signal, and if I get it hot driving a dummy load, it goes up to 17mV. This seems a pretty reasonable method, but I have not done a lot of this kind of work. Opinions, please?
I strongly suggest you to follow the manufacturer/designer instructions
Because if you do different, will probably waste your time when the designer already did that for you.
If you have suspections the designer was mad/drunk/idiot, then go trying.... i do think that, at the end, you will conclude the same adjustment he suggested...he has just saved your time.
The designer has the same brain you have and i have...no one is really much more clever than other...the big difference are skills about some subject and the time the designer wasted testing, watching the spectrum analiser, waveform monitor, distortion monitors, temperature meters, current meters, output power meters and so on.... if he concluded this is the best way to do with that particular amplifier... then it is correct...if he is not mad/drunk/idiot.
I do not think a mad/drunk/idiot is able to build amplifiers... they usually take the charge to criticise others because this is often easier.
regards,
Carlos
Because if you do different, will probably waste your time when the designer already did that for you.
If you have suspections the designer was mad/drunk/idiot, then go trying.... i do think that, at the end, you will conclude the same adjustment he suggested...he has just saved your time.
The designer has the same brain you have and i have...no one is really much more clever than other...the big difference are skills about some subject and the time the designer wasted testing, watching the spectrum analiser, waveform monitor, distortion monitors, temperature meters, current meters, output power meters and so on.... if he concluded this is the best way to do with that particular amplifier... then it is correct...if he is not mad/drunk/idiot.
I do not think a mad/drunk/idiot is able to build amplifiers... they usually take the charge to criticise others because this is often easier.
regards,
Carlos
Try to find OLD Philips databooks somewhere. The old Philips transistor datasheets are very good, but for some reason newer Philips/NXP datasheets contain less and less usable information. This was already so in the valve age; if you compare Philips EZ80 datasheets from different years you will see that the older the datasheet, the better.
Anyway, in a bipolar transistor the main changes are a reduction of VBE and an increase of hFE with increasing temperature.
About MOSFETs: these have various long-term degradation effects, such as hot carrier degradation, negative-bias temperature instability (NBTI), positive-bias temperature instability (PBTI) and time-dependent dielectric breakdown (TDDB). Modern deep submicron MOSFETs in integrated circuits have a lifetime in the order of two months when you put the full supply voltage between drain and source and apply about half supply to the gate. Reducing drain-source voltage a bit (or increasing channel length somewhat) helps a lot.
I'm sure the same physical effects occur in power MOSFETs, but I don't know what kind of lifetime power MOSFETs have under worst-case conditions, nor how quickly it improves by derating the voltages.
You can make a class (A)B amplifier that is not sensitive to temperature differences between the power devices and a temperature sensor by adding a class AB bias loop, thereby also getting rid of variations in distortion due to the dynamics of the music. As far as I know, the class AB bias loop was invented in 1976 by Frans Tol and professor Johan H. Huijsing ('Monolithic operational amplifier with improved HF behavior', IEEE Journal of Solid-State Circuits, vol. SC-II, no. 2, April 1976). My own audio amplifier with class AB bias loop is described in Electronics World February 1996, pages 140...143. It still works very well today, in fact it is playing in the background at the moment.
Last edited:
Very many class-B or AB amps in production have very shaky control of the quiescent current of the output stage. Variations in that current can have a big effect on the audible and measured distortion.
The typical Vbe multiplier-output transistor loop just doesn't track that well without a lot of careful work. As such, many amps have quiescent current that rises and falls with ambient temperature and internally generated heat.
One way around this is the kind of matching one finds in an IC. Another way is an explicit bias control loop. Seevinck has a classic article on this.
Beta is also a strong function of temperature, and increasing beta will very often increase the loop gain of the amplifier, especially in-band, which dramatically drops the distortion at all levels, particularly high output power levels.
Akitika GT-101
The typical Vbe multiplier-output transistor loop just doesn't track that well without a lot of careful work. As such, many amps have quiescent current that rises and falls with ambient temperature and internally generated heat.
One way around this is the kind of matching one finds in an IC. Another way is an explicit bias control loop. Seevinck has a classic article on this.
Beta is also a strong function of temperature, and increasing beta will very often increase the loop gain of the amplifier, especially in-band, which dramatically drops the distortion at all levels, particularly high output power levels.
Akitika GT-101
There are many threads with problems adjusting ADCOM GFA5XX bias.
Nelson Pass occasionally drops in here to suggest the appropriate settings (they're his design originally) but sometimes the matter is
that the model has several revisions in the setting. I know of four revisions (issues) of the GFA 555, for example, and all are set differently.
Are you confident that the manual issue matches your revision/model?
big heatsinks take time to reach thermal equilibrium so play some music, get the heat-sinks warmed up and then take measurements, 30 minutes of music is enough time i think....
you can download the scheme at hifi-engine if you have not done it yet...
i will set the output trannie bias current to something like 50mA, so that at test points P2 and P2, you should read something like 11millivolts....do this several times if the idle current is still going up.....
you can download the scheme at hifi-engine if you have not done it yet...
i will set the output trannie bias current to something like 50mA, so that at test points P2 and P2, you should read something like 11millivolts....do this several times if the idle current is still going up.....
Hmm, yes I'm pretty sure the schematic is the right one. It's a GFA-535 Mark I. Interestingly, looking at the manual for the GFA-535 Mark II, it says to warm the amp for 3-5 minutes, adjust for 7mV, then run it at 20W for 10 minutes, and finally let it run but cool to quiescent, then readjust to 7mV. This seems in line with what I discovered with the Mark I.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.