Good morning,
I was recently given two rack mounted professional amplifiers made by DB technologies, model is MPA1004. They have two channels, 500w each into a 4ohm speaker.
They have inside sixteen mosfets, 8x 2SJ162 and 8x 2SK1058, which are quite rare to find nowadays.
After fixing the pcb, replacing the bad resistors and mosfets, It is time for adjustment, but surprisingly this amplifier has two trimpots for each channel, one for current and one for offset. I never met anything like this before, and I can't find online any info about this and the current needed.
Does anyone have an idea on how these two should be adjusted and where to measure for the calibration?
Regards.
I was recently given two rack mounted professional amplifiers made by DB technologies, model is MPA1004. They have two channels, 500w each into a 4ohm speaker.
They have inside sixteen mosfets, 8x 2SJ162 and 8x 2SK1058, which are quite rare to find nowadays.
After fixing the pcb, replacing the bad resistors and mosfets, It is time for adjustment, but surprisingly this amplifier has two trimpots for each channel, one for current and one for offset. I never met anything like this before, and I can't find online any info about this and the current needed.
Does anyone have an idea on how these two should be adjusted and where to measure for the calibration?
Regards.
Offset: measure the output DC voltage and turn the trimming potmeter until the DC voltage is as close as it gets to zero.
I'm not sure about the other, as I don't know the target quiescent current.
I'm not sure about the other, as I don't know the target quiescent current.
Thank you, that makes much sense. Now I just need an idea for the current trimmer, but sadly there is no manual anywhere.
Are there rail fuses on the +/- 82v rails? You should easily be able to find a sensible bias point empirically. If you remove a rail fuse and put an ammeter in series with the fuse socket, how much current do you see? The Hitachi HMA-7500 recommends (from memory) around 200mA per rail (it has much lower rail voltages than here, but likely a bigger heatsink, as your amp is designed for forced cooling). Bob Cordell's Hafler DH-200 version runs around 400mA (needs larger heatsinks). This amp is derived from the Hitachi apps book, as are hundreds of Lateral MOSFET amps from the '80s and '90s. The C-Audio SR-606 is similar (I think the SR-404 has a similar amount of output FETs). The H+H V500 is another very similar amp (they could almost be cousins...).
Were you to have a dummy load + FFT analyser, you could look at distortion at various bias points and come to a decent compromise, whereby amp doesn't run too hot, but crossover distortion is low at modest levels. If you don't have this equipment, I think that a bias of somewhere between 100-200mA per rail will be fine. If you bias it too high, the fans will kick in when audio level is low / muted. If the fans are working correctly, you could probably bias it really high, but the benefit of having so much bias will likely be offset against fan noise.
Were you to have a dummy load + FFT analyser, you could look at distortion at various bias points and come to a decent compromise, whereby amp doesn't run too hot, but crossover distortion is low at modest levels. If you don't have this equipment, I think that a bias of somewhere between 100-200mA per rail will be fine. If you bias it too high, the fans will kick in when audio level is low / muted. If the fans are working correctly, you could probably bias it really high, but the benefit of having so much bias will likely be offset against fan noise.
Talking about the rails, yes each channel has its own fuse for + and - rail, so that is a very convenient testing point.
I noticed quite a discrepancy in the rail voltage measurement though, since these amps were rated for 220vac and now we have the new 240v standard. The rail voltage is now +-86 volts, which is 6 volts higher than the rated voltage of the main capacitors (80v 4700uF). I'm not sure if it is a good idea keeping these capacitors or replacing them with 100v rated ones. I know caps have some tolerance, but at the cost of longevity.
I noticed quite a discrepancy in the rail voltage measurement though, since these amps were rated for 220vac and now we have the new 240v standard. The rail voltage is now +-86 volts, which is 6 volts higher than the rated voltage of the main capacitors (80v 4700uF). I'm not sure if it is a good idea keeping these capacitors or replacing them with 100v rated ones. I know caps have some tolerance, but at the cost of longevity.
The schematic suggests that you should have +/-82V, so for an unregulated supply 86V is pretty much within tolerance. I cannot believe that an OEM would have fitted 80V caps when they knew that the rails would be 82V? I have seen 50V caps explode when run on 55V for long periods. It is not a chance that I would like to take with my own projects. I cannot believe that an OEM would take risks like this... It has run like this for many years, so maybe they knew that the cap maker's tolerance was generous and they are safe. But it does not sit well with me.
Who knows, for saving money somewhere and granting that after a while people would have to pay their labs to re-service the amps.. 😆
I have seen at least 6 of these amps, and in most of them, if not all, the main caps were inflated, in one case even leaked. The capacitance of all of them greatly reduced, from 4700uF to 3000 or even 2000 in some, probably caused by working for so long beyond the rated voltage.
But it's just a theory.
Your theory has been proven if you've seen 6 of them. The cap that exploded when run at 55V instead of 50V was a 6,800uF type. It made the most almighty bang, leaving fine fibres all over the workshop. It took ages to clean up.
Did you take a look at the H+H V500? It's attached here. Methinks the designers of this amp were no strangers to the V500...
Attachments
yes, there are some similarities. At least on the v500 they use 120v caps 😆
I will attach another section of the amp, the audio input. Here a CD4052 multiplexer is used to switch between bridge and stereo mode, quite an inventive solution. Have you seen something similar before?
Attachments
The 4000-series CMOS switches have been around for a long time and they are seen in a lot of products (cassette decks would be an example. Some people replace them with relays, as they think the lower noise + THD will give better sound...this is quite funny, in a format that offers 60dB SNR without noise reduction!!)
You could argue that although a mechanical relay will offer better noise + distortion performance when compared to a solid-state switch, the CMOS switch has no moving parts and should last longer. If the rest of the amp were designed for long term reliability, this makes the choice of under-rated capacitors even stranger!
(personally, I have found mechanical relays to be reliable. They tend to wear out most when used on outputs, due to arcing over time - but small signal / sealed ones tend to be highly reliable).
NB - if you go near the CMOS switches, be sure to wear a wrist strap.
You could argue that although a mechanical relay will offer better noise + distortion performance when compared to a solid-state switch, the CMOS switch has no moving parts and should last longer. If the rest of the amp were designed for long term reliability, this makes the choice of under-rated capacitors even stranger!
(personally, I have found mechanical relays to be reliable. They tend to wear out most when used on outputs, due to arcing over time - but small signal / sealed ones tend to be highly reliable).
NB - if you go near the CMOS switches, be sure to wear a wrist strap.
I think 100ma per channel is probably about right. The more MOSFETs in parallel higher the transconductance which gives the opportunity to have lower bias than the standard '100ma per mosfet'
I have many lateral mosfet amplifiers with similar in design ..C Audio SR and RA, HH VX and Rauch P-120. The Rauch uses four mosfets per channel. (two on each side of rails) I measured 100ma at mains input so the bias is running approx 25ma per mosfet. I measured the distortion at 20khz and it was at 0.0062% at 100mV input 8 ohm load.
HH VX150 has also two mosfets on each rail and I can bias them very low. No visible crossover distortion on scope and still very low distortion. Higher bias around 200ma per channel actually increases distortion and heat. (runs hot!)
C Audio SR606-707 have 8 paralleled mosfets per rail and I have the bias at 200ma per channel (25ma per mosfet) and that is fine with low very low distortion. Gets only slightly warm with the fans running on idle. .
I have many lateral mosfet amplifiers with similar in design ..C Audio SR and RA, HH VX and Rauch P-120. The Rauch uses four mosfets per channel. (two on each side of rails) I measured 100ma at mains input so the bias is running approx 25ma per mosfet. I measured the distortion at 20khz and it was at 0.0062% at 100mV input 8 ohm load.
HH VX150 has also two mosfets on each rail and I can bias them very low. No visible crossover distortion on scope and still very low distortion. Higher bias around 200ma per channel actually increases distortion and heat. (runs hot!)
C Audio SR606-707 have 8 paralleled mosfets per rail and I have the bias at 200ma per channel (25ma per mosfet) and that is fine with low very low distortion. Gets only slightly warm with the fans running on idle. .
New patient arrived. Missing mosfet brackets, lid, and with bulging main caps here as well. I confirm the usage of 80v caps despite different pcb version 😒
don't mind my temporary kitchen setup.. lab too cold in winter.
don't mind my temporary kitchen setup.. lab too cold in winter.
If you were to bias that above amp at the claimed 100ma per MOSFET then you would be idling 800ma overall (Nearly 1amp of idling bias drawn from mains) The heatsink will need to be bigger!
This is why I suggested 100ma let channel would be better and I bet your distortion will still be low due to the high negative feedback these amps employ.
This is why I suggested 100ma let channel would be better and I bet your distortion will still be low due to the high negative feedback these amps employ.
This amp is equiped with fans: heatsink is big enough.
Lateral mosfets are immune to thermal runaway and easy to bias - the natural bias point is where the temperature coefficient is zero. This point is around 100mA per mosfet and 50mA should be acceptable as well. Lower than that crossover distortion will increase.
More info here.
Lateral mosfets are immune to thermal runaway and easy to bias - the natural bias point is where the temperature coefficient is zero. This point is around 100mA per mosfet and 50mA should be acceptable as well. Lower than that crossover distortion will increase.
More info here.
Something I will definitely try is measure the "factory" bias current adjustment in the new unit I received, since the trimmers have not been moved, and see what order of magnitude it is.
But before that, I checked all of the mosfets with this instrument and all of them have a Drain-Source resistance of 0.2-0.4 ohms, except this one in particular, which is very strange...
But before that, I checked all of the mosfets with this instrument and all of them have a Drain-Source resistance of 0.2-0.4 ohms, except this one in particular, which is very strange...
How Impossible would be replacing all those lateral mosfets with some normal hexfets (irfp240/9240)? Other than a different pinout, it would require changes in the circuit as well...
Despite scouting the entirety of the stores here I couldn't find a single lateral mosfet that isn't fake or doesn't cost 20€ each...
Despite scouting the entirety of the stores here I couldn't find a single lateral mosfet that isn't fake or doesn't cost 20€ each...