I have an ARC Sd135 amplifier that uses the now defunkt Thermal track output transistors. these are failure prone devices and despite the MFG's attempts to fix the issues with them, the product failed. The Thermal track NJL3281 is just a standard MJL3281 with an added internal diode. what fails in these is the thing that is supposed to make them better, the added Internal Thermal Tracking diodes! and that is exactly what has failed in the amp I am working on. thankfully it failed without doing any damage. often times that is not the case. Often times the output devices overheat and short. In my amp the output devices are still intact, but the diodes inside have gone wonky as that is the best way to describe what happens. leaky is maybe a better term. In this amp the bias is WAY out of range and jumps around all over the place. Looking to save this amp, I need to come up with a solution.
I disconnected the string of internal diodes and I experimented with an external string of 1n4148 diodes. 5 diodes gets the bias close to being in range but I would need a way to mount 5 diodes in contact with the output devices or the heatsink. Looking at the Thermal Track whitepaper they show an example with a standard VBE Multiplier. I like that approach better as it is adjustable and I did some experimenting by breadboarding up the VBE circuit and attaching the A06 transistor between 2 of the output devices on the massive heatsink they are mounted on. and with some tweaking I was able to get the amp to idle pretty close to factory spec. which is about 100ma per output device averaged across all the devices. 20mv on the .2ohm emitter resistors. I was able to get the amp to idle for 1 hour at about 18.6mv per device. BUT, I ran the amp at 1/3 power for 5 minutes and then back to idle and i observed that the bias was quite high. about 30mv per device. I may need to move the bias transistor to the middle of the heatsink and maybe come up with a better clamp for it.
*
But before I start too far down that rabbit hole. I thought to throw it to you guys for some (TV show House ) diagnostic style input! I don't want to redesign the whole amp. But I can make a small daughter board with the VBE Multiplier on it that could be adapted to the amp.
Thoughts, feedback, Input?
I disconnected the string of internal diodes and I experimented with an external string of 1n4148 diodes. 5 diodes gets the bias close to being in range but I would need a way to mount 5 diodes in contact with the output devices or the heatsink. Looking at the Thermal Track whitepaper they show an example with a standard VBE Multiplier. I like that approach better as it is adjustable and I did some experimenting by breadboarding up the VBE circuit and attaching the A06 transistor between 2 of the output devices on the massive heatsink they are mounted on. and with some tweaking I was able to get the amp to idle pretty close to factory spec. which is about 100ma per output device averaged across all the devices. 20mv on the .2ohm emitter resistors. I was able to get the amp to idle for 1 hour at about 18.6mv per device. BUT, I ran the amp at 1/3 power for 5 minutes and then back to idle and i observed that the bias was quite high. about 30mv per device. I may need to move the bias transistor to the middle of the heatsink and maybe come up with a better clamp for it.
*
But before I start too far down that rabbit hole. I thought to throw it to you guys for some (TV show House ) diagnostic style input! I don't want to redesign the whole amp. But I can make a small daughter board with the VBE Multiplier on it that could be adapted to the amp.
Thoughts, feedback, Input?
Attachments
I did that a couple of years ago. Vbe transistor, 2SC3421, between the collectors of Q5 and Q6. Tapped a hole in the heatsink close to one of the drivers and bobs your uncle. Worked like a charm.
I really don`t know why the good folks at AR didn`t suggest that as a perfect solution to the problem!.
R
I really don`t know why the good folks at AR didn`t suggest that as a perfect solution to the problem!.
R
Having worked there at ARC for 6 years I understand why. they just didn't have the engineering resources for projects like this.
Do you have any more info, schematic for what you did? a Vbe transistor between the collectors of Q5 and Q6 is interesting. what did you do with the thermal track diode section? etc
Do you have any more info, schematic for what you did? a Vbe transistor between the collectors of Q5 and Q6 is interesting. what did you do with the thermal track diode section? etc
Hmmm. This nagged me a bit during the night. I actually gave you a totally wrong description. I don`t have a picture right now, but there should be to zero R links attaching both sides of the output devices to the string of diodes. I merely lifted those links, effectively disconnecting the ThermalTrak diodes, and inserted the Vbe transistor in those holes. The schematic says 1,65VDC on each side, so I just used that voltage as a reference for the Vbe resistor network. One resistor C to B, and resistor in series with a Bourns trimmer B to E to set the standing current.
The amp is actually in my Norwegian abode, so I can present pictures and actual values for the Vbe in a couple of weeks.
Sorry for the lapse yesterday.
R
The amp is actually in my Norwegian abode, so I can present pictures and actual values for the Vbe in a couple of weeks.
Sorry for the lapse yesterday.
R
I would also like to add that ARC lack of willingness/ability to service ThermalTrack utilicing amps reflects poorly on their reputation as THE HIGH END company. Then again I remembered a comment made in the Absolute Sound ages ago. "In order to get help you have to put yourself in the posistion of a sinner, because their equipment are never at fault". Absolute Sound, on the other hand, frequently appeared to be more high strung than Sheldon Cooper in The Big Bang so who knows where the truth lies. Not me, thats for sure.
R
R
That makes much more sense. disconnecting the diode string and putting the VBE multiplier in place. I think i just need to tweak the resistor value a bit and see if i can get the bias servo to respond just slightly faster/track better.
.
arc went through a period for a while there where they were pretty light on engineering staff. the company had been bought and sold a time or two and things were pretty rough there. OH the stories i could tell would make your head spin! they were all but gone a year ago. and i mean like days PAST dead when they were rescued from the bank. the new owners are taking the right steps to get the company back on its feet. they have hired several people for the engineering staff and things are looking up.
.
arc went through a period for a while there where they were pretty light on engineering staff. the company had been bought and sold a time or two and things were pretty rough there. OH the stories i could tell would make your head spin! they were all but gone a year ago. and i mean like days PAST dead when they were rescued from the bank. the new owners are taking the right steps to get the company back on its feet. they have hired several people for the engineering staff and things are looking up.
Hehe, lucky you. You mentioned servo bias. There is only a DC servo, no bias servo,
I actually did the same with a HD220 a few years ago. That one had two problems. First the TT transistors had run wild and killed the mains transformer, it got replaced at a high cost and then it died again. So I redesigned the Vbe and had a toroid transformer special made for it. Customer happy, problem solved.
R
I actually did the same with a HD220 a few years ago. That one had two problems. First the TT transistors had run wild and killed the mains transformer, it got replaced at a high cost and then it died again. So I redesigned the Vbe and had a toroid transformer special made for it. Customer happy, problem solved.
R
correct, but some articles i have read call the VBE-x ciruit a servo...sorry for the confusion....the HD220 DOES have a servo however to correct for any DC offset. Interestingly enough i have two MASSIVE R-core transformers that were for the HD220 prototype! I had a pair of broken HD220 amps that i traded away. wish i had kept them now.
.
The Whitepaper for the TT devices shows a very similar circuit to the ARC SD135 which is no surprise. they often took designs right out of the design guide for various semiconductor devices and then tweaked them to their liking. it uses a 2.2K Cb and a 330 and a pot with no value shown. I used a 1K pot
.
The Whitepaper for the TT devices shows a very similar circuit to the ARC SD135 which is no surprise. they often took designs right out of the design guide for various semiconductor devices and then tweaked them to their liking. it uses a 2.2K Cb and a 330 and a pot with no value shown. I used a 1K pot
Reading Rod Elliots Amplifier Design guide He mentions "
“It is also quite possible to make the resistance too large, so the bias servo will amplify the temperature change too much. This will cause the amplifier's quiescent current to fall as it gets hotter.”
.
so this has me curious and I want to understand more what he is trying to say here. right now the bias changes very slowly. it takes about 15-20 minutes from off to slowly climb to a stable bias current. If i make the resistors larger, will that speed up that climb? make bigger changes per degree of temp?
“It is also quite possible to make the resistance too large, so the bias servo will amplify the temperature change too much. This will cause the amplifier's quiescent current to fall as it gets hotter.”
.
so this has me curious and I want to understand more what he is trying to say here. right now the bias changes very slowly. it takes about 15-20 minutes from off to slowly climb to a stable bias current. If i make the resistors larger, will that speed up that climb? make bigger changes per degree of temp?
The whole point of a Vbe is to control the standing current, Eventually you will get to a point where it slowly wobbles up and down around a value of your choice.
It`s actually possible to predict this to a very high degree, but it involves all data about the semis used, AND rather advanced calculus. Which I don`t master so I use tried and tested values.
Seeing as I have extremely many different and expensive amps with the same problem, most other designers lack the same skill.
R
It`s actually possible to predict this to a very high degree, but it involves all data about the semis used, AND rather advanced calculus. Which I don`t master so I use tried and tested values.
Seeing as I have extremely many different and expensive amps with the same problem, most other designers lack the same skill.
R
It shouldn't be overly difficult to add a Vbe multiplier for the bias spreader. That's probably what I'd do. You'll have to poke a hole in the heat sink so you can mount it.
Alternatively, you might be able to tease out which diode was used in the ThermalTrak devices by examining the device models for those. I seem to recall it being a plain power diode.
Tom
Alternatively, you might be able to tease out which diode was used in the ThermalTrak devices by examining the device models for those. I seem to recall it being a plain power diode.
Tom
