From the datasheets I have the SOA for the 3055 and TIP35C come close, but TIP35C loses out above its SOA break point.
A plot of the two graphs (these may not be 100% accurate as the data was taken from the datasheets: these are approximates) is shown with a 5 ohm+5j reactive load, assuming a 28V swing (50W into 8 ohms). What these reactive load graphs do not show is the frequency at which the elliptical load line applies, since the speaker's inductance will give a phase dependent on frequency. And, as the frequency increases an increasing impedance, so reducing current even if the phase shift is higher. The TIP35 datasheet suggests that a signal somewhere between 100Hz and 1kHz will be fast enough to use a slightly higher SOA rating. But again the normal SOA is claimed to be "single pulse" and as a signal traversing the ellipse will be likely to be repetitive the equivalent power for a cycle really needs to be considered.
In short, the TIP35 ought to manage 50W on +/-35V, but certainly as wg_ski says, no higher.
A plot of the two graphs (these may not be 100% accurate as the data was taken from the datasheets: these are approximates) is shown with a 5 ohm+5j reactive load, assuming a 28V swing (50W into 8 ohms). What these reactive load graphs do not show is the frequency at which the elliptical load line applies, since the speaker's inductance will give a phase dependent on frequency. And, as the frequency increases an increasing impedance, so reducing current even if the phase shift is higher. The TIP35 datasheet suggests that a signal somewhere between 100Hz and 1kHz will be fast enough to use a slightly higher SOA rating. But again the normal SOA is claimed to be "single pulse" and as a signal traversing the ellipse will be likely to be repetitive the equivalent power for a cycle really needs to be considered.
In short, the TIP35 ought to manage 50W on +/-35V, but certainly as wg_ski says, no higher.
Attachments
Rod also suggest not more than +/-35vDC (unloaded) for TIP35/36C but as you know P3A is capable of 90watts into 4ohm & 70watts into 8ohm. Never heard of any failure using these transistors in P3A, otherwise rod wouldn't recommend them because he run a business.
Last edited:
That seems fair enough. You would get the TIP35C driving 4 ohm loads better than a 3055. (I generally regarded the 3055 as a 6A transistor, neve really tried to push it higher. Even getting it to do 6A is a slight challenge. There are things like the 3771/3772 for that. Although the datasheet for those has not changed (200kHz???) I suspect the ft is higher these days. )
Just another point on the SOA graph - looking at the overlap of the elliptical (A.C.) load line against the TIP35 (D.C.) the length of time the SOA curve is exceeded is only 0.565 radians (out of 2pi) or just under 10% of the cycle. That would only be a 1ms pulse equivalent at 100Hz. When the SOA limit is higher, just over the load line in fact. So practically, TIP35 is OK I suspect for +/-35V. (Is that where the 35 came from😉)
Just another point on the SOA graph - looking at the overlap of the elliptical (A.C.) load line against the TIP35 (D.C.) the length of time the SOA curve is exceeded is only 0.565 radians (out of 2pi) or just under 10% of the cycle. That would only be a 1ms pulse equivalent at 100Hz. When the SOA limit is higher, just over the load line in fact. So practically, TIP35 is OK I suspect for +/-35V. (Is that where the 35 came from😉)
Last edited:
Yes, tip35c is great for low impedance load if you use multiple parallel transistors. But here i don't want that much power, 40-50w/ch(8R) only. I mean a cheap but GOOD quasi amp for my friends or for myself. Another unfinished project i have is Elvee's Circlophone which is also Quasi. I have mjl21194(TO-264) for that, same one pair. For complementary amps, c5200/a1943, my first choice.
Regards
Regards
Hi, Circlophone uses same sex output transistors that's why i said that. But you're right.
Regards
Regards
The MJ15022/24 are good, robust, pretty fast power transistors.
The power control world has moved on to IGBTs and MOSFETs; that's where the innovation is. You'll need a different topology for these.
The power control world has moved on to IGBTs and MOSFETs; that's where the innovation is. You'll need a different topology for these.
The power control world doesn’t mess with linear amplifiers. Except for the occasional low power LDO regulator that runs alongside a higher power DC converter of some kind. The innovation really isn’t in the converter circuits themselves, but in the processor controlling everything.
While IGBTs could be used in a QC or circlo, why would you want to when a pair of 15024’s does that better? By “different topology” I suppose you mean class D or one of its variants. The old topologies will work, but just give you ho-hum who-cares results. And maybe a pile of blown IGBTs for your trouble.
While IGBTs could be used in a QC or circlo, why would you want to when a pair of 15024’s does that better? By “different topology” I suppose you mean class D or one of its variants. The old topologies will work, but just give you ho-hum who-cares results. And maybe a pile of blown IGBTs for your trouble.
I am not advocating class D (perish the thought). I mean a QC with a MOSFET or IGBT cannot use this Baxandall diode topology.
I think the latest devices are better than old BJTs.
I think the latest devices are better than old BJTs.
Last edited:
I don’t think of the Baxandall diode as a topology, but as a circuit technique used with a QC to improve it. I think of topology as in higher level - circlo vs. Lin, SE vs. complementary - that sort of thing.
By “latest devices” I hope you mean BJTs. Yeah, they are “better”, but with the right topology Old And Slow are “good enough”. Newer (well they are not that new now, LAPTs have been around a while) devices can be used to get distortion down to ppm levels and slew rates into the 1000V/us range. I don’t believe one has to to make a good sounding amplifier. I try not to touch “Latest” mosfets with a 10 foot pole. One often has to, and then the application must be rigorously tested before being trusted. Its going back to the old “How lucky do you feel, Punk?” days when SOA was not fully understood unfortunately. If I wanted to make a linear amplifier with mosfets I’d be going straight for the Exicons or the old IRFP240.
By “latest devices” I hope you mean BJTs. Yeah, they are “better”, but with the right topology Old And Slow are “good enough”. Newer (well they are not that new now, LAPTs have been around a while) devices can be used to get distortion down to ppm levels and slew rates into the 1000V/us range. I don’t believe one has to to make a good sounding amplifier. I try not to touch “Latest” mosfets with a 10 foot pole. One often has to, and then the application must be rigorously tested before being trusted. Its going back to the old “How lucky do you feel, Punk?” days when SOA was not fully understood unfortunately. If I wanted to make a linear amplifier with mosfets I’d be going straight for the Exicons or the old IRFP240.
It is possible to make a high fidelity amplifier using IGBT's but such amplifiers are rare. Fet & BJTs are more popular.
They don’t like being used in linear amplifiers any more that modern switching-optimized mosfets do. Data sheets may show 600 watts of thermally limited SOA but its liar liar pants on fire. After you’ve blown up a handful of $11 apiece IGBTs for ”no reason” you’ll go back to mosfets. The old ones. IRFP240 isn’t totally free from second breakdown either, but it’s real world capability is comparable to the MJ15024.
Better yet, just use a pair of 15024’s - with the Baxandall diode. Use high speed sustained beta drivers. That is far more critical - and you can always find somebody’s version of them in stock, somewhere.
Better yet, just use a pair of 15024’s - with the Baxandall diode. Use high speed sustained beta drivers. That is far more critical - and you can always find somebody’s version of them in stock, somewhere.
If I recall correctly Toshiba gave John Linsley Hood some IGBT's to see if he could make an amp out of them.
I do not recall any results of that - so I suspect nothing particularlly exciting emerged.
I do not recall any results of that - so I suspect nothing particularlly exciting emerged.
What is a LAPT? According to Wikipedia, it is either a Latin American Poker Tour or some protein that is related to cancer, but I assume you mean something else.
Large area parallel transistor?
A term to describe a process with lots of emitter and base contacts which "ring emitter" and other geometries allow.
Generally keeping the base resistance low and SOA up also with high ft. Generally meaning lots of parallel emitters, I take it, with possible emitter ballast resistor layer.
A term to describe a process with lots of emitter and base contacts which "ring emitter" and other geometries allow.
Generally keeping the base resistance low and SOA up also with high ft. Generally meaning lots of parallel emitters, I take it, with possible emitter ballast resistor layer.
I suspect a small fire. Exciting perhaps, but not in the way you would want. If IGBTs had good real world SOA everybody would be using them.
LAPT = large area parallel transistor. But I have seen more recent Sanken data sheets referring to them as “linear amplifier power transistor”. Perhaps they are using that term for any of their types which exhibit the high fT and high SOA, regardless if they are the ballasted multi-emitter variety.
Just looked at a datasheet for a 230W 50A 600V IGBT. Could not manage 20W over 20V. In fact the datasheet mentioned it was designed for one type of switching regulator and not to be used for any other application.