I have been troubleshooting for a long time and recently took to studying the finer details of transistor design and refinement. Many output driver transistors look as if they were intended for beta independent circuits, the beta curves are pretty bad looking humpty dumpties.
How about these ? They are sort of rare, low volume things, I figure showing them may boost sales well enough to prolong the product life. I hope they keep making them. I bought 15 pairs for evaluation.
How about these ? They are sort of rare, low volume things, I figure showing them may boost sales well enough to prolong the product life. I hope they keep making them. I bought 15 pairs for evaluation.
Last edited:
I have them in mind for an Szikali driver pair for these. Aim is 200watt with an option for doubling the output transistors. I do get nit-picky but figure the easier I can get them linear without need for correction, the end result may be better yet. Since I am not making a million of them I can ignore $3.50 output transistors instead of $1.50 - $2.50.
It is a try to replace a 30 year old mosfet design with four hard to get $15 each lateral mosfets.
Above all, a learning experience.
It is a try to replace a 30 year old mosfet design with four hard to get $15 each lateral mosfets.
Above all, a learning experience.
Last edited:
They are a little small for drivers in a 200 watt application. 100 watt is more appropriate.
These may be on the way out - a lot of similar devices like the A1507/c3902 pair have already bit the dust. Similar devices in SMD packages will likely still be available for years, but if you want isolated TO 126's that are NOT the BD139 type you probably ought to stock up while you can. It is only a matter of time before they all disappear.
Flatness of the hfe curve is far more important on the driver transistor than it is on the output. Nice to have in both places, but more critical on the driver.
These may be on the way out - a lot of similar devices like the A1507/c3902 pair have already bit the dust. Similar devices in SMD packages will likely still be available for years, but if you want isolated TO 126's that are NOT the BD139 type you probably ought to stock up while you can. It is only a matter of time before they all disappear.
Flatness of the hfe curve is far more important on the driver transistor than it is on the output. Nice to have in both places, but more critical on the driver.
I worked for a company called ILP Electronics many years ago and joined them at a time when they were using output transistors designed for an 8 ohm resistive load. When I showed the MD/CEO the load curves relating to a real loudspeaker (+/- 60 degrees phase shift) and saying that the design needed three pairs of devices in parallel, I hit a brick wall. Luckily, I'm now free to design what I like, how I like.
Not in the US. Digikey no reference, newark references the NTE39 with all the specs left off the datasheet.
digikey has actual stock of ksa1220ay and ksc2690ay in Minnesota. Farnell has theirs in the UK only.
20 watt 155 mhz drivers, 200 ma soa at 70 v. cool. I just bought four 2n5682 in the TO39 metal can for drivers at $1.88 each. These are ECB instead of EBC and the heat sink would be cheaper since it has a flat surface. I may buy some, they would be cheaper than TO39 heat sinks.
Last edited:
Using those as a Darlington pair wouldn't even reach the bottom of the SOA would it ? +- 65v supply, about 4.5amps /100/100 ~ 450 micro amps.
This is my first scratch design, I always did production line and customer site troubleshooting. Now I have some time to dig in deeper.
This is my first scratch design, I always did production line and customer site troubleshooting. Now I have some time to dig in deeper.
Last edited:
The cool thing about a non-darlington package unlike TIP107/147 is that the heat of the output transistor doesn't automatically flow to the driver transistor . So if a driver is not put on the output transistor heat sink, even if hooked up as a darlington there is less tendency to thermally run away than a real darlington.
There was a tendency to darlington output PA amps in the 80's, BDsomething, that Sakis criticized because they didn't hold up in service. He ran a repair shop in Piraeus, presumably getting the repairs from all the beach bars on the islands. So he should know.
I think 4 Mhz Ft output transistors are fine for classical music, because I have one channel that sounds good with MJ15003 output. The channel that sounds good has 1970 RCA 5 digit TO5 drivers, presumably with two digit Ft. I repaired the other channel with 3 mhz Ft drivers, TIP41/42, and the high frequency response on bells and top octave piano is not very good. MJE15031/32 or 28/29 drivers could be good enough with 30 mhz Ft, but I haven't tried it yet. Those are TO220.
For a little less board space there is BD139/140, but the only ones stocked in the US i've found are fairchild, which specifies no Ft at all. The Phillips datasheet had 190 mhz specified, but I can't buy them. The 2sa1220/2sc2690 from fairchild proves they have the fast die, but they want more money for them than a BD139/140. These 2sa1220/2sc2690 are $.48 now in eachs and $.23 in hundreds.
I have a pair of boards from Jiny designed for darlington output TIP102/107. They are tiny and will fit in my 3.5" tall chassis, but I'm suspicious of the darlington outputs for the reasons Sakis posted. Jiny used BC546/556 as predrivers with 270 ohms rail resistors. I'm thinking of building them with MJE15028/29 ootput transistors, and upping the drive current by changing the predriver resistors to 150 ohm. In that case a heat sink on predriver might be required, as on my TO39 transistors, or on BD139/140 or more ideally 2sa1220/2690. 2 digit Ft minimum seems to matter on predriver, driver, and VAS. The predrivers need to be small because JINY fit the pads in a 1/4" circle
So thanks wavewhipper for highlighting these. I tend to use newark because they are close and ship to me in 1-2 days, but they weren't stocking or highlighting the KSC parts in my searches.
There was a tendency to darlington output PA amps in the 80's, BDsomething, that Sakis criticized because they didn't hold up in service. He ran a repair shop in Piraeus, presumably getting the repairs from all the beach bars on the islands. So he should know.
I think 4 Mhz Ft output transistors are fine for classical music, because I have one channel that sounds good with MJ15003 output. The channel that sounds good has 1970 RCA 5 digit TO5 drivers, presumably with two digit Ft. I repaired the other channel with 3 mhz Ft drivers, TIP41/42, and the high frequency response on bells and top octave piano is not very good. MJE15031/32 or 28/29 drivers could be good enough with 30 mhz Ft, but I haven't tried it yet. Those are TO220.
For a little less board space there is BD139/140, but the only ones stocked in the US i've found are fairchild, which specifies no Ft at all. The Phillips datasheet had 190 mhz specified, but I can't buy them. The 2sa1220/2sc2690 from fairchild proves they have the fast die, but they want more money for them than a BD139/140. These 2sa1220/2sc2690 are $.48 now in eachs and $.23 in hundreds.
I have a pair of boards from Jiny designed for darlington output TIP102/107. They are tiny and will fit in my 3.5" tall chassis, but I'm suspicious of the darlington outputs for the reasons Sakis posted. Jiny used BC546/556 as predrivers with 270 ohms rail resistors. I'm thinking of building them with MJE15028/29 ootput transistors, and upping the drive current by changing the predriver resistors to 150 ohm. In that case a heat sink on predriver might be required, as on my TO39 transistors, or on BD139/140 or more ideally 2sa1220/2690. 2 digit Ft minimum seems to matter on predriver, driver, and VAS. The predrivers need to be small because JINY fit the pads in a 1/4" circle
So thanks wavewhipper for highlighting these. I tend to use newark because they are close and ship to me in 1-2 days, but they weren't stocking or highlighting the KSC parts in my searches.
Arrow was the only US distributor that had both in stock at the time I ordered.
I was going on thoughts that the more linear the system is without gain, the better it will perform with gain. Some of the grass on the waveform could be from the feedback system hunting for equilibrium. As long as the output noise is in low millivolts who cares ?
But I just wanted to see what this polishing the polish stuff is worth.
I was going on thoughts that the more linear the system is without gain, the better it will perform with gain. Some of the grass on the waveform could be from the feedback system hunting for equilibrium. As long as the output noise is in low millivolts who cares ?
But I just wanted to see what this polishing the polish stuff is worth.
MJE 340/350 don't have the nice flat beta curve, regardless of who makes them. They do however work quite well for the 1st stage of an output triple. Triples are often used at high voltages, where you want more than a 160 volt part. With a triple, flat hfe isnt anywhere near as obvious as it is with an EF2.
I had been wondering if those humpy beta curves do best in the mid volume range but start showing themselves at low (average listening levels) and when drunk enough to turn the thing up past the last 3/4 of volume range.
I think actually, the trick to beta independent circuits is to not push them to the curve's limit and the voltage amp mode will hold it's performance ability. In that case fewer but better transistors may be the answer to the need for extra voltage gain stages . And sure, a loose design would be easier to sub repair parts in the future. What I'm doing may be difficult to find a repair substitute for in the future.
A few days later and I'm still learning. Who has some of that NZT-48 stuff ?
Oh well, it's getting late.
I think actually, the trick to beta independent circuits is to not push them to the curve's limit and the voltage amp mode will hold it's performance ability. In that case fewer but better transistors may be the answer to the need for extra voltage gain stages . And sure, a loose design would be easier to sub repair parts in the future. What I'm doing may be difficult to find a repair substitute for in the future.
A few days later and I'm still learning. Who has some of that NZT-48 stuff ?
Oh well, it's getting late.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.