I’ve built amplifiers where 230 volts isn’t enough, even 300 not being enough. For the predriver or the VAS. I have store bought units in the same power range, ie, Crest CA18. The parts they use go from the ridiculous to the sublime - MJE13007 and MJE5282. Yes, it uses an enhanced VAS and EF3. Wouldn‘t be worth a crap if it didn’t.
An extra 20 MHz won’t buy much. Your application may be incredibly sensitive where the difference could be measured in some way but differences on that order are usually buried in the floor set by build technique. The extra 500 mA probably doesn’t mean much either by itself, as drivers or predrivers are never run up to the current rating. If you are only running an EF2 stage, you want to limit normal operation of the drivers to the flat part of the hFE curve. In an EF3 the predriver always will, and it’s not as critical for the driver to do so. Bigger drivers can be used then anyway. The current at which gain starts dropping may or may not even be any different between a 1 and 1.5A part. Go by that curve and the SOA for part selection when you have a choice of what to use.
Great, I’ll grab some of both. I recently found the importance of high gain. Normal recommend replacement for C1451 is the KSC3503, but only D ranking available. Found some C2911S that make a better replacement, I found that the 3503 doesn’t always make the best replacement for those.
I still am curious what you guys think, given an amplifier circuit where both work well within their limits, since I have the C3298 and A1306 now is there any reason to search out the C3884 and A1859? Does that extra 20 MHz make that much of a difference?
Dan
Sorry, I did not see this message so I had asked the question again. Should’ve gone to the next page lol.
I guess I don’t know how to properly rate these devices for use. How could 300v rating not be sufficient for an amp with high voltage rails of +/-95v? Where could it see more than 300v across two of its legs?
Dan
Last edited:
CA18 has +/-160 volt rails…. I’m working on some that go to +/-185, and +/-140 and 127 have already been done. +/-95 is the QSC USA1310, which used C3298/A1306B drivers. One pair driving 8 MJ15024/5’s, and the only two things that ever go out on those are capacitors and the thermistor in the bias stack. So the old ones were pretty rugged compared to new ones.
Oh I must have misread the schematic. Holy smokes, hate to be zapped by that. I guess I have worked on amps needing those high voltage parts, but just haven’t needed them. I fixed a small stack of Yamaha PC9501N amplifiers, I can’t find the rail voltage in the schematic, but they use 200v rated caps and the drivers are C2333 and A1009, which are 350v - 500v rated transistors. Luckily they’re not needed that often.
The Crest CA18 is a beautiful amplifier, if you want heavy iron they don’t get much better, maybe the CA20. I have a CA9 picked up (broken) for $20 at an electronics convention, looks nice on the inside. Someday I’ll get to it.
Dan
My bad, very bad OCD activates when it comes to matching devices. I know they would work together but when I see those, not same markings on transistors.... man it's hard to describe.
One last question on this subject, don’t mean to dwell. So fast speed is considered better. The C4883 is 120 MHz, I thought the complement A1859a was 120 as well, but it’s only 60 MHz. So when you have a NPN and PNP pair working together to drive outputs how does that work? Say one pair being 60 MHz and 120 MHz vs another pair that’s 100 MHz and 100 MHz. How are those compared?
Dan
Dan
The requirement is for the driver to be significantly faster than the output.
The 120 MHz is probably specsmsnship anyway. For that type of transistor the NPN is often faster, and it does depend on Vce and Ic. In the region where speed actually matters (near zero crossing) vce is high and fT is as high as it’s going to get.
The 120 MHz is probably specsmsnship anyway. For that type of transistor the NPN is often faster, and it does depend on Vce and Ic. In the region where speed actually matters (near zero crossing) vce is high and fT is as high as it’s going to get.
Okay, so the PNP only being 60 MHz isn’t the biggest of deals since it’s likely faster than most modern outputs anyways? I have to think on what you said for a bit to fully grasp it, just woke up.
Would you say there is a superior scenario, having drivers that are equally fast, ie 100 MHz and 100 MHz. Or having drivers where the PNP is half the speed such as in the C4883/A1859?
Dan
Of course it would be a “superior” solution, but is it superior enough to even be able to tell? In a properly executed amplifier, using modern techniques and parts, what you ultimately end i with depends on other factors besides small differences in characteristic between similar parts intended for the same service.
And they may be closer than you are led to believe anyway. You’d have to test them independently under the same conditions to get an apples to apples. Sanken’s been known to test fT at higher vce than typical to get higher numbers. Toshiba (and way back when, Sanyo) has been known to publish an average for the pair. As I said, specsmanship.
And they may be closer than you are led to believe anyway. You’d have to test them independently under the same conditions to get an apples to apples. Sanken’s been known to test fT at higher vce than typical to get higher numbers. Toshiba (and way back when, Sanyo) has been known to publish an average for the pair. As I said, specsmanship.
That is a very good point, I am guessing that telling the differences apart will be extremely hard if at all. Well, I will be building two of the Wolverine amplifiers, and I do intend on using one pair of each in the amplifiers. I don’t have a bunch of fancy test equipment, but I can measure it on the QA403 and then of course my ears, but even they are what matter most they are also the least reliable. Thank you for answering my questions.
Dan