Heads up Part 2...do not buy BEL, particularly if abroad, as they are so reputed that junk is re-marked as BEL at times, as the supply to civil market is not a priority for them.
Unless it is a really reputed seller who stands by his material.
BEL are more into military radios, radars, tank gun controllers, complete submarine and naval suites and so on! I have some 2N2866 RF transistors, purchased from their shop on St. Mark's Road, now shifted to their plant in the suburbs.
CDIL are good, or as wg_ski says, put MJ or 3773, those will be safer as higher SOA ratings exist for those.
Another thing, local may be better, simply because the unit cost landed at your door may be the same or less than CDIL shipped, as those will cost about $ 4 each...I think you said 50 cents each and $35 shipped, what about Customs duty?
Yes, if genuine, the CDIL parts are worth buying in large quantities, the shipment may be the same for a larger weight, so think of getting more parts to save on unit cost.
You can always get different parts from the same supplier, to satisfy your curiosity about CDIL.
Once again, I emphasize that I have no ties to CDIL.
Unless it is a really reputed seller who stands by his material.
BEL are more into military radios, radars, tank gun controllers, complete submarine and naval suites and so on! I have some 2N2866 RF transistors, purchased from their shop on St. Mark's Road, now shifted to their plant in the suburbs.
CDIL are good, or as wg_ski says, put MJ or 3773, those will be safer as higher SOA ratings exist for those.
Another thing, local may be better, simply because the unit cost landed at your door may be the same or less than CDIL shipped, as those will cost about $ 4 each...I think you said 50 cents each and $35 shipped, what about Customs duty?
Yes, if genuine, the CDIL parts are worth buying in large quantities, the shipment may be the same for a larger weight, so think of getting more parts to save on unit cost.
You can always get different parts from the same supplier, to satisfy your curiosity about CDIL.
Once again, I emphasize that I have no ties to CDIL.
Last edited:
Edit (after time up)...
2N3866 RF output transistor, not 2866.
BEL also makes TV transmitters, up to the RF tubes, and most X Ray machines here in India have BEL tubes as emitters.
Like I said, transistors are a small part of their business, it is one of a very few silicon refinery to military equipment operations in the world.
St. Mark's Road is in Bangalore / Bengaluru, now the third largest city in India, a famous tech hub.
Also, a defense manufacturing and education hub.
2N3866 RF output transistor, not 2866.
BEL also makes TV transmitters, up to the RF tubes, and most X Ray machines here in India have BEL tubes as emitters.
Like I said, transistors are a small part of their business, it is one of a very few silicon refinery to military equipment operations in the world.
St. Mark's Road is in Bangalore / Bengaluru, now the third largest city in India, a famous tech hub.
Also, a defense manufacturing and education hub.
Their Small signal parts are fine & i use them too but my experience with CDIL power devices wasn't pleasant. Their Quality control is not that good & that reflects in their price.
https://www.diyaudio.com/community/threads/manufacturing-fault-cdil-2n3055.382143/
https://www.diyaudio.com/community/threads/manufacturing-fault-cdil-2n3055.382143/
Ooops! Especially from you, Indianajo, as a musician and a Hammond organist, I would have expected to know the sonic difference between the original tube amps and some random SS amps. I'd rather go for the heat ;-).
Best regards!
Yep - and when doing work with Mil electronics you are just about never given a choice in replacement parts - you use what was used before or on the (very short) alternate part list. The assemblies are almost always proprietary and the manufacturers won't sell you just the $1.00 IC - but will be happy to sell you another 50K$ VXI card to replace the faulty one. And even if the replacement part is available - semiconductors in particular - there is no guarantee that a process change at the manufacturer won't cause the replacement to fail. Identical data sheet specifications, not identical in reality.
When I joined this forum. I asked a question. Say the amp maker gives me a list of possible transistors that I can use. How do I pick the best one from that list. Another question I asked is say I have a bag of transistors how do I pick the best one to use. Got some posts calling me an idiot. But no clear answers. A few months down the road.
I have figured out a few things.
a. Build a test gig to measure the noise of the transistors. There are some good links and schematics google throws up.
b. Build it and see how it sounds / works on the scope.
I have also been hitting the junk yard and picking up 2nd hand heat sinks, Toroidal transformers and junk boards.
Also have been buying the same part from different vendors esp when there is a huge variation in the price.
I like to pull the parts off high end boards and measure them. Nobody seems to be using the To92 package numbers I see popping up here. They all seem to use different transistors. Wonder why they pick the transistors they pick. i.e. big name guys. Like bose, Akai, Techniques etc. None of them use stuff like BC547, 2n2222 etc.
Yesterday a large batch of transistors arrived from CDIL
What I found was that the same transistor number does not match on the component tester.
First step is to put every transistor purchased thru the component tester and log the numbers.
I have found that if its a good batch / maker / vendor. I always buy like 25 or 50 numbers. As I know they will all be from the same batch.
Then test each one. The good guys seem to spit out numbers very close to each other.
Like all the 2n3055Hs I got measure the same. The NPN To92 philips measure the same. Very little variation part to party.
The one where I see a large variation in the numbers spit out. I tend to doubt. i.e. in my mind uniformity somehow relates to quality. Yesterdays batch of BD139 and BD140 from CDIL did not show much uniformity.
Next phase is to make the circuit to measure transistor noise.
This is one of the links I found looking at other options also. And still going thru old posts on this forum. On how to measure noise.
http://www.dicks-website.eu/low_noise_amp_part3/part3.html
I have figured out a few things.
a. Build a test gig to measure the noise of the transistors. There are some good links and schematics google throws up.
b. Build it and see how it sounds / works on the scope.
I have also been hitting the junk yard and picking up 2nd hand heat sinks, Toroidal transformers and junk boards.
Also have been buying the same part from different vendors esp when there is a huge variation in the price.
I like to pull the parts off high end boards and measure them. Nobody seems to be using the To92 package numbers I see popping up here. They all seem to use different transistors. Wonder why they pick the transistors they pick. i.e. big name guys. Like bose, Akai, Techniques etc. None of them use stuff like BC547, 2n2222 etc.
Yesterday a large batch of transistors arrived from CDIL
What I found was that the same transistor number does not match on the component tester.
First step is to put every transistor purchased thru the component tester and log the numbers.
I have found that if its a good batch / maker / vendor. I always buy like 25 or 50 numbers. As I know they will all be from the same batch.
Then test each one. The good guys seem to spit out numbers very close to each other.
Like all the 2n3055Hs I got measure the same. The NPN To92 philips measure the same. Very little variation part to party.
The one where I see a large variation in the numbers spit out. I tend to doubt. i.e. in my mind uniformity somehow relates to quality. Yesterdays batch of BD139 and BD140 from CDIL did not show much uniformity.
Next phase is to make the circuit to measure transistor noise.
This is one of the links I found looking at other options also. And still going thru old posts on this forum. On how to measure noise.
http://www.dicks-website.eu/low_noise_amp_part3/part3.html
I also found exactly this in transistorized Hammond organs. Literally any SS device is inhouse numbered, even simple small diodes. I guess they just want you to buy $$$ replacement parts from them instead of soldering stock JEDEC items into the organs.
Best regards!
Guilty as charged 😉
The Hometaxial 3055 (the 60´s and early 70´s ones) were VERY rugged; later Epitaxials were much weaker and failed in the Real World.
I used Hometaxials by the thousands from 1969 until 2006-2008 when the supply dried out.
Last ones available in production/commercial quantities were India made UR-USHA
That’s the other elephant in the room concerning the old hometaxials. Real world ruggedness. The domestic and many Jap epitaxials and triple diffused would take the data sheet maximums, but not much more. Full power to 30 or 40 volts, then fall like a stone. The homo 3055 showed full power to 60 volts. The reality is it would take even more than that, if you could run any given batch at higher Vce. So if you were comparing an MJ15003 to 3055H at 60 volts, the 3055H would actually have it beat. The current gain was another story - but in a guitar amp you can use 60 watt drivers so you don’t have to beat those to death.
Maybe they just weren’t available in your country, and maybe still not. But if you could have, maybe you should have switched to Toshiba triple diffused parts, like the 2SC3281 when your supply of 3055H ran out. For ruggedness they gave the MJ15024 a run for its money. Unfortunately all good things must end and the C5200 was one step forward and three steps back. The old Toshiba D424 was also real world rugged (often passing the 2N6259 second breakdown test spec), but by 2008 those were basically obsoleted by the C3281. On’s version is as good, but they resort to a bigger die to do it. But if you we’re having trouble getting On’s for a reasonable amount in your country then, it’s only worse now.
Maybe they just weren’t available in your country, and maybe still not. But if you could have, maybe you should have switched to Toshiba triple diffused parts, like the 2SC3281 when your supply of 3055H ran out. For ruggedness they gave the MJ15024 a run for its money. Unfortunately all good things must end and the C5200 was one step forward and three steps back. The old Toshiba D424 was also real world rugged (often passing the 2N6259 second breakdown test spec), but by 2008 those were basically obsoleted by the C3281. On’s version is as good, but they resort to a bigger die to do it. But if you we’re having trouble getting On’s for a reasonable amount in your country then, it’s only worse now.
Yes, Motorola and On-Semi devices were my go-to's for reliability. I used TONs of 2SD424 and 2SD555 (and their compliments) in service work. I have found that the MJ15024/MJ15025 were solid, and used those in Carver equipment. Generally everything actually. Having said that, there was the horrible MJ15015 / MJ15016. But if you read the data sheet, they were listed as "economy parts". That should have been warning enough to avoid using them.
Lots of good devices, but it didn't pay to stock a huge assortment. By stocking what was actually needed, I was able to match them more closely when needed. Higher quantities of one number also lowers the price of the part.
The original 2N3055 was designed to be bullet proof. They were in industrial power supplies, and had low gain-bandwidth products so they didn't oscillate. That process had reliability as the prime characteristic, but it was at odds with the needs of audio reproduction.
Lots of good devices, but it didn't pay to stock a huge assortment. By stocking what was actually needed, I was able to match them more closely when needed. Higher quantities of one number also lowers the price of the part.
The original 2N3055 was designed to be bullet proof. They were in industrial power supplies, and had low gain-bandwidth products so they didn't oscillate. That process had reliability as the prime characteristic, but it was at odds with the needs of audio reproduction.
In its day it had a great feature, the 15A continuous rating, but in practice this wasn't very usable as the base current needed to properly saturate it at 15A was perhaps 3A or more, requiring another large power transistor to drive it - so no advantage over a Darlington really. Its collector current was probably overstated: the ?current crowding? or whatever mechanism drops its performance well before the 15A rating. Some datasheets give no data above 8A for instance, which is a more realistic rating really...
Yep - typical beta of 5 at 15 amps. This is also true of MJ15015/6. Which is pretty pathetic for an audio transistor intended to run up to +/-60 volts. Those were always out of there so fast it would make your head spin.
Back in the TO-3 day, you needed to stock exactly 3 types. D424, D555, and MJ15024. And the PNPs of course. When plastic came along, the C2922 and C3281. What driver types to use depended on what MCM had on hand that particular day/week. There were a dozen or more, pretty much interchangeable.
Back in the TO-3 day, you needed to stock exactly 3 types. D424, D555, and MJ15024. And the PNPs of course. When plastic came along, the C2922 and C3281. What driver types to use depended on what MCM had on hand that particular day/week. There were a dozen or more, pretty much interchangeable.
That is exactly right!
I did stock those, and a few others. In addition the plastic types you mentioned and a few other numbers. Too many skews!
I did stock those, and a few others. In addition the plastic types you mentioned and a few other numbers. Too many skews!
I actually used pretty much everything under the sun to build and experiment with. But that short list is pretty much the only ones I ever used to fix Other People’s Amplifiers after about 1982. That and the occasional D1047, but I often got asked if I could put in something tougher, and that would be the 3281. Little stuff that used PNP Ge’s, TO-66’s or TO-220’s for outputs was on a case by case basis.
Yup, the odd ones were always case-by-case.
I used buckets of 2SD1047 & 2SB817 and others. I tried to keep to something similar so I didn't run into oscillation issues. My source bought direct from Japan, life was wonderful until he closed shop. Then it was pretty much a Digikey (or similar) experience to be certain I was getting real whatever.
Ever try the On-Semi MJW0281A / MJW0302A? Those were fantastic devices!
I used buckets of 2SD1047 & 2SB817 and others. I tried to keep to something similar so I didn't run into oscillation issues. My source bought direct from Japan, life was wonderful until he closed shop. Then it was pretty much a Digikey (or similar) experience to be certain I was getting real whatever.
Ever try the On-Semi MJW0281A / MJW0302A? Those were fantastic devices!
Not yet - I haven’t had to repair a TO-3P amp in over 10 years. And the last one of those was a car amp. Last few actual stereos to come through here were those case by case jobs - one had PNP Ge outputs and that wasn’t even what was wrong with it (they were running the MPSA63 driver hot as blazes, so both got changed with D41K). It’s really tapered off, and I don’t advertise or anything. Hell I’m not even in the DJ business anymore. But if someone gave me a call out of the blue, with enough warning before the gig I’d find a way to get it done. And my heyday of rebuilding old vintage monster amps and PA gear is pretty well over too. All of mine are in good shape, and nobody’s bringing me Phase Linears to work on anymore.
All of my recent builds have been in two camps. The big stuff and the serious stuff all uses the TO-264’s. MJL21194 or 3281 from On, or C5200 from Toshiba. Got enough of the 21194 and 15024 to rebuild anything along the way, but new construction uses the flatpacks. The other camp is all smaller stuff using stone knives and bear skins from the “stash”. 80 volt epitaxials. TIPs. Germanium. Output caps. Some around here might be shocked to hear how GOOD only 40 watts using obsolete devices can sound if you make a few minor tweaks to a 50 year old design and apply all of Self’s cardinal rules concerning EM compatiblity. THIS is why all the interest in the old 2N3055 stuff.
All of my recent builds have been in two camps. The big stuff and the serious stuff all uses the TO-264’s. MJL21194 or 3281 from On, or C5200 from Toshiba. Got enough of the 21194 and 15024 to rebuild anything along the way, but new construction uses the flatpacks. The other camp is all smaller stuff using stone knives and bear skins from the “stash”. 80 volt epitaxials. TIPs. Germanium. Output caps. Some around here might be shocked to hear how GOOD only 40 watts using obsolete devices can sound if you make a few minor tweaks to a 50 year old design and apply all of Self’s cardinal rules concerning EM compatiblity. THIS is why all the interest in the old 2N3055 stuff.
Well, most of my work is in rebuilding older amplifiers and updating them with new part types and circuit techniques. Yes, Self is bang on. I get paid well to improve performance and get great results in general. Some designs won't ever come up to the levels normally possible, so i turn those down unless it is a straight repair. I always measure "as found" and completed performance. What I do is not found on the net anywhere.
My biggest problem are the hacked units I get from bad techs (who shouldn't be in business) to guys who copy internet based "upgrades". They make a huge mess out of what was good equipment. Those things suck up time and I tend top lose money on them. Customers don't tell you up front as a rule and you have to discover these things.
Those MJW transistors are NLA. I used them in designs and some kits, like the Symasym where they improved performance. It is too bad they didn't continue to offer those parts, they were really amazing, matching very closely as well.
My biggest problem are the hacked units I get from bad techs (who shouldn't be in business) to guys who copy internet based "upgrades". They make a huge mess out of what was good equipment. Those things suck up time and I tend top lose money on them. Customers don't tell you up front as a rule and you have to discover these things.
Those MJW transistors are NLA. I used them in designs and some kits, like the Symasym where they improved performance. It is too bad they didn't continue to offer those parts, they were really amazing, matching very closely as well.
I still saw NJW0281/0302 last I was on Mouser a few days ago. One was out of stock, as was the 4302 I was looking for, as was the MJE5731A. No Biggie, the actual NEED my be two years out after I finish building the new house. But I’ve learned to get things while I can as they actually may BE gone in a year or two. I’ve got this giant “bucket list” amp on the horizon (actually, a bank of four of them), and I’m considering the 4281/4302 as drivers instead of the 2119x. I want the higher Vceo to lower the leakage at high temp. Same with the VAS - it was MJE15035, but I want the higher SOA at 350V the slower device offers. It’s darlington‘ ed anyway so it’s input characteristic is dominated by a smaller device. I got it working beautifully on +/-140V rails 10 years ago, and I’ve recently wound up some transformers to go to +/-165 on next gen. Switched voltages, of course. No way to deal with the heat otherwise. This is what I meant by “big stuff”, and why TO-3Ps hold no special fascination when the 264 will run cooler.