People have looked at this before, Nigel.
Before Mr Self's book, manufacturers were striving for what Mr Self calls "blameless". Let me remind you of the German designed and made Mega 1 power amp from Linear Systems - I suppose you could call that one something of a blameless amp, or at least, what preceeded the entire concept. There, you had a low noise BC414B as the pass transistor for a cascode in conjunction with a BFT28A.
This is, in fact, exactly what you are asking now. Just take a look at the supplied schematic. It has its faults, for example no input stage degeneration, but overall, it was way ahead of its time when it appeared circa 1978.
Several samples were built locally in its day, and frankly, it sounded very, very good. Convincing definition and detail.
Today, I wouldn't settle for just two 15,000 uF caps feeding both channels, but then, I don't design or make commercially orientated units. In other words, let the poor thing breathe and damn the cost.
Please note the AA132 germanium diode D5 - the authors claim it reduces settling time to about 1/10 of what it would be without it. A solution not often seen.
Note that it used what I consider to be the unsung hero power transistor pair of all time, TI's BD 249C/250C. Very crafty and deceptive decives, just looking at their specs would show that it isn't really suited for the German amp, given that they are rated at just 100V. But also 25A continuous, and all of 40A in peaks.
You might think they are the European version of TI's TIP 35C/36C, and in many ways they are, but for one key difference - their switching times are WAY better than you'd expect from a 3 MHz device, and about 50% faster than TIP 35C/36C. Which makes them very good output devices.
Before Mr Self's book, manufacturers were striving for what Mr Self calls "blameless". Let me remind you of the German designed and made Mega 1 power amp from Linear Systems - I suppose you could call that one something of a blameless amp, or at least, what preceeded the entire concept. There, you had a low noise BC414B as the pass transistor for a cascode in conjunction with a BFT28A.
This is, in fact, exactly what you are asking now. Just take a look at the supplied schematic. It has its faults, for example no input stage degeneration, but overall, it was way ahead of its time when it appeared circa 1978.
Several samples were built locally in its day, and frankly, it sounded very, very good. Convincing definition and detail.
Today, I wouldn't settle for just two 15,000 uF caps feeding both channels, but then, I don't design or make commercially orientated units. In other words, let the poor thing breathe and damn the cost.
Please note the AA132 germanium diode D5 - the authors claim it reduces settling time to about 1/10 of what it would be without it. A solution not often seen.
Note that it used what I consider to be the unsung hero power transistor pair of all time, TI's BD 249C/250C. Very crafty and deceptive decives, just looking at their specs would show that it isn't really suited for the German amp, given that they are rated at just 100V. But also 25A continuous, and all of 40A in peaks.
You might think they are the European version of TI's TIP 35C/36C, and in many ways they are, but for one key difference - their switching times are WAY better than you'd expect from a 3 MHz device, and about 50% faster than TIP 35C/36C. Which makes them very good output devices.
Last edited:
Hi Nigel,
Starting a new thread on audio small-signal and power transistors that are good and available is a good idea. This is a continuing problem, and one that I will have to address in the Second Edition of my book.
Cheers,
Bob
Come on, Nige, I don't really give you advice, I'm simply sharing my experience. I honestly see no point in not doing so, if it can help you.
And you must admit, you did really like the BF 720/721 pair, didn't you? Damn good transistors.
Thanks Dvv . Now everyone lets list how good and opinion over fact is fine . Price even better . For example I think KSP 94 ( MPSA 94 ? ) at 8000 piece at 3 cents a piece is worth a thought from Farnell . RS have them at 15 cents in 10 's . Last chance saloon for TO92 400 V PNP . Ideal to make a high voltage triple . After that SMD .
High voltage home testing of Vce . How do I make machine for $10 to do it correctly ? Is it a Variac and a 10 M resistor WO8 rectifier and capacitor ?
BF 720/721 , life saver . I suspect it is the T0126 BF device in a new pakage ?
High voltage home testing of Vce . How do I make machine for $10 to do it correctly ? Is it a Variac and a 10 M resistor WO8 rectifier and capacitor ?
BF 720/721 , life saver . I suspect it is the T0126 BF device in a new pakage ?
Not here, Nige, over there.
But no, I think 720/721 are different, wider bandwidth transistors, rated not as "min. 60 MHz", but as "100 MHz".
My current preference is the well known 2SC3503/2SA1381.
Not exactly small, but any reason not to use these?
There are Sanyo video drivers like 2SA17772SC4623 that are faster but more Cob. Mostly moot since apparently unotainable. I would be interested if anyone knows a source.
I have basic models of the 3503/1381 if you need them.
Best wishes
David
I hope so, Bob. It's downright amazing at what we collectively know. Also, it's good to sometimes face reality and take a step back in some humility.
Not long ago, I asked myself - if cascode is good, and if parallel is good, why not parallelled cascode? Oh, I was feeling so proud of myself just then. Then, a day or two later, I discover that Nelson Pass had used excactly that way back in the early 90ies in his Threshold 400A/4000 amps and probably elsewhere too.
Oh well, such is life.
Hi Bob , On that thread Steven asked if someone would keep a list of the transistors . I wondered if you might show your list every month ? I know you plan to publish the list in your next book so thought it would be OK to ask ?
Nigel
Hi Nigel,
My list right now is just static, and I'm not sure when I'll get to that part of the second edition. As I learn more about new transistors available, I'll be happy to post that information, but I think it is pretty far down the road. Unfortunately, I'm no further along than anyone else on that thread, at this point - maybe behind many.
Cheers,
Bob
Continuing with this theme: http://www.diyaudio.com/forums/soli...self-wants-your-opinions-182.html#post3606113
Douglas, on page 265 you show the schematic of the "load invariant" design. I note, however, that you used a deeply compromised single-slope SOA protection circuit that should really never be used at all.
This is because it is exceedingly inefficient and wasteful of perfectly useable transistor SOA.
I have run a SICE simulation (Simetrix) of you arrangement with a properly calibrated simulation fixure (see attachment) and found that the maximum available current from your pair of output devices is slightly less than 6A, as predicted by simple calculation.
Additionally, the amplifier cannot be used with supply rails greater than +/-24V (~35W@8Ohms) as the SOA protection would be activated even with no load connected to the output. Moreover, the supply rails would have to be reduced even further if ambient temperatures in excess of 25 degrees Celcius are anticipated in the vicinity of the SOA protection transistors.
Douglas, on page 265 you show the schematic of the "load invariant" design. I note, however, that you used a deeply compromised single-slope SOA protection circuit that should really never be used at all.
This is because it is exceedingly inefficient and wasteful of perfectly useable transistor SOA.
I have run a SICE simulation (Simetrix) of you arrangement with a properly calibrated simulation fixure (see attachment) and found that the maximum available current from your pair of output devices is slightly less than 6A, as predicted by simple calculation.
Additionally, the amplifier cannot be used with supply rails greater than +/-24V (~35W@8Ohms) as the SOA protection would be activated even with no load connected to the output. Moreover, the supply rails would have to be reduced even further if ambient temperatures in excess of 25 degrees Celcius are anticipated in the vicinity of the SOA protection transistors.
Last edited:
Good info . I leave the protection devices out anyway . If using TO3 on flying leads and design for disaster it's a 20 minute repair and rather good fun . I like capacitor coupling as it is the ultimate DC offset protection for real world people who smoke other than cigarettes . Humble Rotel amplifiers seldom gave trouble with 4AT fuses . The student customers had many new fuses on the house . 99% of the time that was all and we are not talking once . Knowing that a 6AT3 fuse will not blow after months on our workshop kettle ( 10 A ) I realized fuses allow vastly greater currents than we imagine . I think using the same transistors as Rotel and 4AT it should be OK . For 6 amps use two sets . I don't know if it was just me ? Quad 405 with blown fuses always meant transistors also . Rotel cheap and bulletproof . Ultra careful Quad less so . Lets be frank Quad were excellent as mini PA . Rotel were better albeit at lower voltage output average ( 21 Vrms , just fixed a RA 931 , just fuses ) .
My friend uses Simetrix . He was blowing transistors . The data has to be correct for the simulation to work . He says if a complex load double the current you think the maximum . This meant 50 A for his application . John is a scientist and a very good one ( too good , a bit intense ) . He got it wrong . The big mistake was to make the design able to run speakers . That isn't the company requirement . The device costs £50 000 the last time I asked and there is a waiting list to buy one . It wasn't for sale even . People heard they had one and wanted one . It is for calibration of magnetic field measuring equipment ( Magnetometer ) . Magnetometers are wonderful . How can you measure a static field ? Not easy . Asymmetry and saturation . The rest is a deadly secret .
A DC crowbar I do approve of . Dam the amplifier , save the speakers .
My friend uses Simetrix . He was blowing transistors . The data has to be correct for the simulation to work . He says if a complex load double the current you think the maximum . This meant 50 A for his application . John is a scientist and a very good one ( too good , a bit intense ) . He got it wrong . The big mistake was to make the design able to run speakers . That isn't the company requirement . The device costs £50 000 the last time I asked and there is a waiting list to buy one . It wasn't for sale even . People heard they had one and wanted one . It is for calibration of magnetic field measuring equipment ( Magnetometer ) . Magnetometers are wonderful . How can you measure a static field ? Not easy . Asymmetry and saturation . The rest is a deadly secret .
A DC crowbar I do approve of . Dam the amplifier , save the speakers .
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.