power fets, bias, gain matching & other stuff.
g'day folks this is my first post, I've lurked a little from time to time now heres my first question.
I have a hideous old mosfet power amp come in from an installation. It has been replaced with something far more recent & it isn't going back to the customer.
I've set about sorting the poor ****** out and giveing it a chance at a new life. Its certainly not a commercialy viable repair but from time to time I like a chalenge.
This is an 20 year old amp with mosfet output stage prety well straight out of the hitachi app notes.
4 pairs of fets in the output stage of each channel claiming arround 350 watts a channel. arround 90 volt rails.
Now I don't believe that this could ever be called a HIFI amplifier, its an uggly sucker but we'll see how well it comes up.
The bias currents of the various output devices are all over the place, & i'm aware that mosfet output devices degrade with heat and age so I willl be replacing the entire output stage with new 10p20 & 10n20 devices.
I've had this sort of unit come in before but have always turned them arround as a minimum cost repair, I'm interested to see how well they come up when "fixed properly".
the pcbs have already been reworked with most of the caps replaced and a variety of track repair done
my questions to the forum are.
what is considered to be the suitable bias currents in these units for varoious applications.
How much benifit is there in gain matching the fets in the output stage.
What is the currently considered best method of measuring the gain of power fets out of circuit.
Is there a convienient source of the original hitachi application notes, I have the exicon stuff.
I've had a bit of a search around but can't easily find what I want. I can remember reading a post a while ago that had some stuff concerning gain matching but It dosn't seem to pop up easily.
cheers chaps
g'day folks this is my first post, I've lurked a little from time to time now heres my first question.
I have a hideous old mosfet power amp come in from an installation. It has been replaced with something far more recent & it isn't going back to the customer.
I've set about sorting the poor ****** out and giveing it a chance at a new life. Its certainly not a commercialy viable repair but from time to time I like a chalenge.
This is an 20 year old amp with mosfet output stage prety well straight out of the hitachi app notes.
4 pairs of fets in the output stage of each channel claiming arround 350 watts a channel. arround 90 volt rails.
Now I don't believe that this could ever be called a HIFI amplifier, its an uggly sucker but we'll see how well it comes up.
The bias currents of the various output devices are all over the place, & i'm aware that mosfet output devices degrade with heat and age so I willl be replacing the entire output stage with new 10p20 & 10n20 devices.
I've had this sort of unit come in before but have always turned them arround as a minimum cost repair, I'm interested to see how well they come up when "fixed properly".
the pcbs have already been reworked with most of the caps replaced and a variety of track repair done
my questions to the forum are.
what is considered to be the suitable bias currents in these units for varoious applications.
How much benifit is there in gain matching the fets in the output stage.
What is the currently considered best method of measuring the gain of power fets out of circuit.
Is there a convienient source of the original hitachi application notes, I have the exicon stuff.
I've had a bit of a search around but can't easily find what I want. I can remember reading a post a while ago that had some stuff concerning gain matching but It dosn't seem to pop up easily.
cheers chaps
"Big mistake. "
Don't have a clue?
I'll help you out a bit. The 10p20 & 10n20 are Exicon part numbers.
http://www.profusionplc.com/static/images/data sheets/ecf10n20.pdf
As we can see from the data sheet, if we bias at or above 200mA the device cannot go into thermal runaway.
"EXICON Mosfets have been designed specifically for high power linear use.
They offer high voltage capability, high slew rate and low distortion, making them the ideal choice for audio amplifier design.
Freedom from secondary breakdown and THERMAL RUNAWAY make them extremely reliable and remove the need for protection circuitry.
These advantages together with wide bandwidth, low drive requirements, and ease of paralleling make it possible for the simple construction of robust amplifiers with excellent sonic characteristics. "
Don't have a clue?
I'll help you out a bit. The 10p20 & 10n20 are Exicon part numbers.
http://www.profusionplc.com/static/images/data sheets/ecf10n20.pdf
As we can see from the data sheet, if we bias at or above 200mA the device cannot go into thermal runaway.
"EXICON Mosfets have been designed specifically for high power linear use.
They offer high voltage capability, high slew rate and low distortion, making them the ideal choice for audio amplifier design.
Freedom from secondary breakdown and THERMAL RUNAWAY make them extremely reliable and remove the need for protection circuitry.
These advantages together with wide bandwidth, low drive requirements, and ease of paralleling make it possible for the simple construction of robust amplifiers with excellent sonic characteristics. "
I am referring to the exicon fets that are intended to replace the hitachi devices.
from what I read they are suposed to be closer spec and more consistant than the old hitachi 's from which tehy are derived.
the reason for the question about bias is thet the bloke who designed this amp has a tendancy to bias things a little hard.
I fixed one of his bipolar designs for a bloke once, he braught it back complaining that one side was running hot & not sounding as clean as the other. The hot side was the side I hadn't touched.
I just biased it up to what I thaught was a fair thing & saw no crossover dist'.
I've seen a number of fet output stages go so far out of spec that the adjacent fet s track so badly you can see a knee in the wave
thanks for the quick reply chaps
from what I read they are suposed to be closer spec and more consistant than the old hitachi 's from which tehy are derived.
the reason for the question about bias is thet the bloke who designed this amp has a tendancy to bias things a little hard.
I fixed one of his bipolar designs for a bloke once, he braught it back complaining that one side was running hot & not sounding as clean as the other. The hot side was the side I hadn't touched.
I just biased it up to what I thaught was a fair thing & saw no crossover dist'.
I've seen a number of fet output stages go so far out of spec that the adjacent fet s track so badly you can see a knee in the wave
thanks for the quick reply chaps
After a solid evenings 
web crawling I may have answered some of my own questions.
there are some good articles at the pas labs site and a couple of others I googled up.
so I'm off to sit down with a cupa & a bicky & see if these articles tell me what I need to know.
any further comments welcomed.
cheers
web crawling I may have answered some of my own questions.there are some good articles at the pas labs site and a couple of others I googled up.
so I'm off to sit down with a cupa & a bicky & see if these articles tell me what I need to know.
any further comments welcomed.
cheers
"do you happen to have an adress where to buy the Exicon's ?"
Profusion plc (UK) sells direct and ships worldwide.
If you have a local Farnell they generally have them and the Semelab/Magnatec BUZ900/BUZ905 as well.
Original Hitachi J49/K134 for $10 each at:
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&category=50593&item=5740965501&rd=1
Call and ask about J56/K176, they have them too.
Profusion plc (UK) sells direct and ships worldwide.
If you have a local Farnell they generally have them and the Semelab/Magnatec BUZ900/BUZ905 as well.
Original Hitachi J49/K134 for $10 each at:
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&category=50593&item=5740965501&rd=1
Call and ask about J56/K176, they have them too.
I had a bit of a read of the articles I've grabbed & I come to the following conclusions.
Gain matching of the output stage is a great idea, but I dont have enough stock to realy select effectively.
the fets I have are from the same batch and are the same devices so I'm a long way ahead of what was loaded when it came to me. one module had a variety of devices loaded the other had two pairs of different devices.
(the rock & roll trogs only replace what is blown )
At this point my efforts are probably better spent adjusting the circuit to suit the execon fets as they differ a little from the hitachi.
I have a current circuit with all the progressive mods which seem to make sense with what I have read.
Seems the heavy bias these units commonly run is probably reasonable somewhere arroung 80 to 100ma per device.
Seems that some of the amp "notes" look more at bias voltage at the gates than measuring the current in the devices as a method of setup.
still interested in experiences of others on bias.
I've been learning quite a bit in the last few days about fet output stages (which is the point of the exercise).
It always amazes me how little you need to know to fix something & get it out the door, its only when you start fiddling & trying to understand that you learn anything.
still interested if anyone can point me to the specs & apps on the old TO3 hitachis.
As far as availability
In the past I have got hitachi replacements from radio spares in aus, not cheap.
The best place I've found for exicons in aust is altronics 10p20s & 10n20s at arround 17 dollars aust retail or arround 14 dollars aust on a trade deal.
cheers
Gain matching of the output stage is a great idea, but I dont have enough stock to realy select effectively.
the fets I have are from the same batch and are the same devices so I'm a long way ahead of what was loaded when it came to me. one module had a variety of devices loaded the other had two pairs of different devices.
(the rock & roll trogs only replace what is blown
)At this point my efforts are probably better spent adjusting the circuit to suit the execon fets as they differ a little from the hitachi.
I have a current circuit with all the progressive mods which seem to make sense with what I have read.
Seems the heavy bias these units commonly run is probably reasonable somewhere arroung 80 to 100ma per device.
Seems that some of the amp "notes" look more at bias voltage at the gates than measuring the current in the devices as a method of setup.
still interested in experiences of others on bias.
I've been learning quite a bit in the last few days about fet output stages (which is the point of the exercise).
It always amazes me how little you need to know to fix something & get it out the door, its only when you start fiddling & trying to understand that you learn anything.
still interested if anyone can point me to the specs & apps on the old TO3 hitachis.
As far as availability
In the past I have got hitachi replacements from radio spares in aus, not cheap.
The best place I've found for exicons in aust is altronics 10p20s & 10n20s at arround 17 dollars aust retail or arround 14 dollars aust on a trade deal.
cheers
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.