I strongly recommend to you: Bob Cordell's 'Designing Audio Power Amplifiers' pp. 319-341. This chapter is up-to-date at a DIY level, and relatively complete.
As far as relay or any other protection is concerned, it is sometimes a necessary evil, in order to save amps and loudspeakers. Small amps can be protected with just a pair of circuit breaker contacts in series with the power supply rails followed by a few hundred uf in capacitance to damp any ringing. But, over +/- 30V or so, they are not enough. It is an 'ideal' protection scheme, however. All of my Parasound designs are +/- 50V to +/- 100V. It is impossible to NOT protect with relatively sophisticated V-I limiting, and not have to pay for expensive speaker damage. We tend to use dedicated IC's that are made in Japan especially for sophisticated circuit protection. I will put up a typical part # when I come across a data sheet.
As far as relay or any other protection is concerned, it is sometimes a necessary evil, in order to save amps and loudspeakers. Small amps can be protected with just a pair of circuit breaker contacts in series with the power supply rails followed by a few hundred uf in capacitance to damp any ringing. But, over +/- 30V or so, they are not enough. It is an 'ideal' protection scheme, however. All of my Parasound designs are +/- 50V to +/- 100V. It is impossible to NOT protect with relatively sophisticated V-I limiting, and not have to pay for expensive speaker damage. We tend to use dedicated IC's that are made in Japan especially for sophisticated circuit protection. I will put up a typical part # when I come across a data sheet.
TA7317 was made by Toshiba and is still widely available.
There is a datasheet in English. No need to translate.
TA7317 Datasheet catalog
TA 7317 P - IC-SCHALTUNG - ICs SAA.. - TCA.. bei reichelt elektronik
Best regards,
Patrick
.
There is a datasheet in English. No need to translate.
TA7317 Datasheet catalog
TA 7317 P - IC-SCHALTUNG - ICs SAA.. - TCA.. bei reichelt elektronik
Best regards,
Patrick
.
Hi,
The fuses will not protect the Amp unless using lateral mosfets and often not even then. Take my word, when the output DC protection cuts in (if it is correctly designed) the Amp's is already a goner. No point protecting it any more.
Well, with no output load the Amp should be ok, but I would not rely on it.
Bob Cordell's book has several options to deal with this problem, have a look.
Ciao T
The fuses will not protect the Amp unless using lateral mosfets and often not even then. Take my word, when the output DC protection cuts in (if it is correctly designed) the Amp's is already a goner. No point protecting it any more.
Well, with no output load the Amp should be ok, but I would not rely on it.
Bob Cordell's book has several options to deal with this problem, have a look.
Ciao T
Hi,
Relay Crowbar (shunt) across the output and a relay based shunt mute have zero effect on measured and subjective performance, because they are simply open circuits when open.
As long you make sure by suitable relay choice that your mute relay engages first you can also tie in other protection.
If we are not very interested in economy I find it best to avoid fold-back current limiters and the like completely and just more and bigger transistors and heatsinks. Of course when attempting to compete in a given market economics are usually more important.
But as DIY'er, why not use 12pairs 2SC5200/2SA1943 for a 150W/8Ohm Amp with 50V Rails?
It reduces distortion at all levels and allows you around 2.16A Iq for optimum Class B Bias with 0.1 Ohm Emitter resistors, so around 36W Class A into 4 Ohm.
Then you can relax SOAR protection requirements considerably and rely on the mute relay to protect the Amp from excessive tomfoolery.
Sure, 216W steady state dissipation per channel is not greatly liked in this day of the "green" aera, but with maybe a John Curl Style Frontend on separate regulated rails and with a pair of Hitatchi lateral Fets as drivers (see: http://www.linearaudio.nl/Miscellaneous/jc.JPG) of our big transistor banks I'm sure it would be one heck of an Amp...
Ciao T
Relay Crowbar (shunt) across the output and a relay based shunt mute have zero effect on measured and subjective performance, because they are simply open circuits when open.
As long you make sure by suitable relay choice that your mute relay engages first you can also tie in other protection.
If we are not very interested in economy I find it best to avoid fold-back current limiters and the like completely and just more and bigger transistors and heatsinks. Of course when attempting to compete in a given market economics are usually more important.
But as DIY'er, why not use 12pairs 2SC5200/2SA1943 for a 150W/8Ohm Amp with 50V Rails?
It reduces distortion at all levels and allows you around 2.16A Iq for optimum Class B Bias with 0.1 Ohm Emitter resistors, so around 36W Class A into 4 Ohm.
Then you can relax SOAR protection requirements considerably and rely on the mute relay to protect the Amp from excessive tomfoolery.
Sure, 216W steady state dissipation per channel is not greatly liked in this day of the "green" aera, but with maybe a John Curl Style Frontend on separate regulated rails and with a pair of Hitatchi lateral Fets as drivers (see: http://www.linearaudio.nl/Miscellaneous/jc.JPG) of our big transistor banks I'm sure it would be one heck of an Amp...
Ciao T
John,
The NEC uPC1237 and the NTE7100 are form, fit and function equivalents to the TA7317. I have used either with the same results.
jan
Thanks Jan.
For the record, effective amp protection that does NOT interfere with sound quality is VERY difficult to do. Most approaches are a compromise, including Parasound's, which is rather elaborate.
At CES, I met up with Bob Stuart, of Meridian. I first met him in London, almost 50 years ago, at a Baxandall lecture, and invited to his house to attend a party. During the party, Bob, with my support, CUT OUT the protection circuits from his new power amp prototype, BECAUSE we could hear them. Bob and I had a serious discussion that night about protection circuits, among other things.
For the record, effective amp protection that does NOT interfere with sound quality is VERY difficult to do. Most approaches are a compromise, including Parasound's, which is rather elaborate.
At CES, I met up with Bob Stuart, of Meridian. I first met him in London, almost 50 years ago, at a Baxandall lecture, and invited to his house to attend a party. During the party, Bob, with my support, CUT OUT the protection circuits from his new power amp prototype, BECAUSE we could hear them. Bob and I had a serious discussion that night about protection circuits, among other things.
Last edited:
39 million Google hits. I am working my way through Cordell and Self. They both go deep into SOA and preventing the amp from doing stupid things by itself. I like the idea of collapsing the bias spreader if it detects over-current like shorted speaker leads. Should not the outputs be robust enough to withstand a short at idle? If so, rapidly muting the input could save the outputs. The amp I am playing with is the little DH-120. Cheap, disposable, and basically a decent amp. The new outputs are 8A L-MOSFETS and the rail fuses remain 4A. They should survive for a few hundred miliseconds.
Sleeping on it, the crowbar worries me as if it false triggered, then damage could be high. Looking at Mouser, many times more contact alloys than I remember. That will take some research. I'll start with simple mute and AC relays as I need them anyway. Better than nothing. I'll keep studying.
Appreciate the tips. Ya know, in the winter time, one could justify a 100W class A amp as you have to heat the house anyway. Just like incandescent bulbs we are using the "waste" heat so they are 100% efficient. I have a 1200W electric heater in my workshop, I could just have a big Pass!
Sleeping on it, the crowbar worries me as if it false triggered, then damage could be high. Looking at Mouser, many times more contact alloys than I remember. That will take some research. I'll start with simple mute and AC relays as I need them anyway. Better than nothing. I'll keep studying.
Appreciate the tips. Ya know, in the winter time, one could justify a 100W class A amp as you have to heat the house anyway. Just like incandescent bulbs we are using the "waste" heat so they are 100% efficient. I have a 1200W electric heater in my workshop, I could just have a big Pass!
Thread topic?
All of these details are related to amp design. The thread is why amps sound different. Different designs as I don't think anyone here believes in magic.
What can we measure by objective means or by inspection that makes deferential amps sound different? My selfish goal is to get schooled so I understand why design differences sound different and then adjust the design or implementation to correlate what we hear with what I am measuring. My measurement could be 100% irrelevant to the sound. Don't know until I try.
The most recent direction on protection is that it can effect performance, and that I did find the limits of a insufficient protection design that is also pointed to as sounding poor. (simple output fuses). When I get it all back together, we will do the same listening tests with new ringers, HCA 1200 and Creek 4120, previous winner, RA 840, and previous loser, DH-120, and see if the simple power supply, mechanical and protection changes changes to the 120 resulted in subjective sound differences, then repeat my contrived pulse test to see if the measurement correlates. See, it all ties together.
Nifty little chip. Does not require extra power supplies. Limited, but better than where I am at. Coffee is ready. Time to research relay contact alloys. I have been "informed" that even the little thump from my Creek that we wake up to in the morning needs to be "fixed". I think it only has a line fuse.
http://relays.te.com/appnotes/app_pdfs/13c3236.pdf
Looks like a basic primer.
In the old T-Bar buss and tag relays, I seem to remember we used two sets of bifurcated contacts, or 4 contacts per line to reduce resistance.
Looks like a basic primer.
In the old T-Bar buss and tag relays, I seem to remember we used two sets of bifurcated contacts, or 4 contacts per line to reduce resistance.
Pretty much confirms, to maintain good contacts you must prevent arcing. So, that means input muting before picking the output relay, and mute before dropping it. It also means the relay could become sacrificed on fault. I thought we used a rhodium alloy but I have not seen reference to it yet. We had to withstand 80 amp arcing as well as millivolt consistency. More searching. It looks like gold-silver-nickel with sufficient redundant contacts is a reasonable approach.
Hi,
Sorry for the continued OY, though it is tvrgeeks thread...
Maybe a moderator would care to separate the amplifier design consideration" part of the thread from the "Sound Quality vs. Measurements"?
Though protection circuitry for example can have dramatic effects on amplifier measured and sonic performance, especially with real speakers attached.
At idle, usually yes.
At max signal...? Maybe. Care to find out?
Yes they do.
As I have commented before, if you use a small signal relay (say NEC UB2) it will mute within 2mS if you use shunt mute. This may be fast enough if you have some reserve in the output stage, check the 10mS single shot SOAR which normally is present in Datasheets.
Honestly, if you use the crowbar only as DC protection and you make sure to trigger mute at the same time, there is no danger.
Use a NEC UB2 relay as mute and a NAIS 10A JW Series relay as crowbar.
The UB2 will mute within 2mS of having power applied and will best manage 1mS (mechanical system limit).
The JW series must operate within 15mS and it's mechanics cannot close faster than 5mS so it will take at 3mS longer to short the output than it will take the UB2 to mute the input, pure mechanics make sure of it.
Moreover, with huge numbers of Amp's (including many that where not mine and did not use my "dual crowbar" circuit) over many years I am hard pressed to think of any instance where DC protection triggered without the Amp already output a lot of DC as result of a shot output...
Ciao T
Sorry for the continued OY, though it is tvrgeeks thread...
Maybe a moderator would care to separate the amplifier design consideration" part of the thread from the "Sound Quality vs. Measurements"?
Though protection circuitry for example can have dramatic effects on amplifier measured and sonic performance, especially with real speakers attached.
At idle, usually yes.
At max signal...? Maybe. Care to find out?
Yes they do.
As I have commented before, if you use a small signal relay (say NEC UB2) it will mute within 2mS if you use shunt mute. This may be fast enough if you have some reserve in the output stage, check the 10mS single shot SOAR which normally is present in Datasheets.
Honestly, if you use the crowbar only as DC protection and you make sure to trigger mute at the same time, there is no danger.
Use a NEC UB2 relay as mute and a NAIS 10A JW Series relay as crowbar.
The UB2 will mute within 2mS of having power applied and will best manage 1mS (mechanical system limit).
The JW series must operate within 15mS and it's mechanics cannot close faster than 5mS so it will take at 3mS longer to short the output than it will take the UB2 to mute the input, pure mechanics make sure of it.
Moreover, with huge numbers of Amp's (including many that where not mine and did not use my "dual crowbar" circuit) over many years I am hard pressed to think of any instance where DC protection triggered without the Amp already output a lot of DC as result of a shot output...
Ciao T
- Status
- Not open for further replies.