PDF mentioned a range of 55-160 (5V, 1A). hFE=237 was a measurement issue or is it so common that measured hfe exceeds the maximum specification? I have never met such outliers myself but someone else did.
MT-200 sankens will out-do many ON/toshiba outputs >SOA @ 60V+
3 pair mt-200 = 5 or more pairs mt-100 or ON's.
They do rank low (<100Hfe) , best for an EF3.
Edit - my ON 0281/0302's are 140+Hfe (both p/n)
- Sanken mt-100's , all over 220. - 35V @ 20ma Ic
All sourced @ Mouser , digikey , and Newark.
OS
Last edited:
True but if the transistor is mismatching at the voltages and currents used by the DMM it will surely also mismatch at higher currents.
Regarding Andrew's comments, about having bought 100 transistors and only getting one pair to match, well he is lucky, I have bought dozens of audio, complementary transistors from Mouser and Farnell, and nothing matches, the NPNs are around 100 and the PNPs start off at 200, going from memory. Tested at various currents.
I even wrote to the manufacturer who promised to "look into it" (ON Semi) , never got anything back from them. But what would they say? "Sorry, our typical hFE curves are fictitious, just look at the "min" settings, and by the way when we say "complementary" we actually just made it up"....
I was talking about Sankens that are sold in my country. They sell various versions with different maximum Vce, with different prices. Because it is official product, I guess they are selling off-spec products. I couldn't find a high quality one. Majority consumers want cheap and do not have ears 😱
My main test gear is my ears. These Sankens do not sound good. I have little experience with high voltage rail and EF3 so I don't know if they will sound good with EF3 😕
Regarding SOA, Toshiba 2SC5200 has been used here in many PA amp power level (around 1000W) so at least for home use I don't think 60V rail is an issue at all.
Find transistors that are specified/advertised as "FOR AUDIO". They have very tight tolerance (and high quality only, no hFE group). For power devices I have Sanken MT-100 but don't remember the model.
For VAS/driver, many people here use 2SA1381 because it looks the best in simulation (if you get the best part as the model of course 😛). It has hfe from 40-320 (C from 40-80, D from 60-120, E from 100-200, F from 160-320). 2SA1360 on the other hand, is specified "for audio" with hfe from 120-240. Lower Cob (1.8pF, 3.1pF for 2SA1381). I don't have hundreds but they are all very close 😎
Last edited:
They are for audio. MJE15028/30/32 and MJE15029/31/33. I have bags full at home and no matches. ON Semi were "surprised" when I wrote to them and promised to investigate, but they never came back. Basically the PNPs are completely outside the range of the NPNs, in real life.
Maybe that is the reason that old designs used NPNs exclusively at the output stage, because they could not match NPNs and PNPs.
And I have seen the same "all NPN" in chip amps and op-amps, eg the LM3886 (excluding reasons like "single supply" for op-amps).
Maybe that is the reason that old designs used NPNs exclusively at the output stage, because they could not match NPNs and PNPs.
And I have seen the same "all NPN" in chip amps and op-amps, eg the LM3886 (excluding reasons like "single supply" for op-amps).
MT-200 "sound bad" ???? What ?
They are specifically made for EF3's ,definite "beta droop" on any EF2's !!
Best audiophile Japanese amps use them , I'm sure they would not if they
"sounded bad" 😕 . Have seen very few fail on 20 year old amps.
You must drive them with MT-100's (EF3 -below). @10A soa -10ms nothing comes close -I'm just using the OPS in the picture for my subwoofer.
BTW - all my sanken's match nearly perfectly. The little ones (mt-100's) have
better gain /better matching than my ON TO-3p outputs.
OS
They are specifically made for EF3's ,definite "beta droop" on any EF2's !!
Best audiophile Japanese amps use them , I'm sure they would not if they
"sounded bad" 😕 . Have seen very few fail on 20 year old amps.
You must drive them with MT-100's (EF3 -below). @10A soa -10ms nothing comes close -I'm just using the OPS in the picture for my subwoofer.
BTW - all my sanken's match nearly perfectly. The little ones (mt-100's) have
better gain /better matching than my ON TO-3p outputs.
OS
Attachments
Last edited:
One of the fastest , lowest cob (driver/outputs) around - 2sa1294/2sc3263.
250 cob / 35mhz Ft - better than a Mje with 3 X the power. 😀 Drivers from hell.
OS
250 cob / 35mhz Ft - better than a Mje with 3 X the power. 😀 Drivers from hell.
OS
Yes , in a class H setup ... where the SOA is split between rail Levels. Never in a
typical Class AB. They suck at 60-70V. Little tiny to-3p ON semi 0281g is about
the same , @ 2.00$ as well.
OS
The Hfe range is marked O (80-160) or R (55-110) on genuine parts. If yours don't carry a grade marking or the grades vary widely from the originals they could well be fake but note that the specified Hfe is measured at much higher current than the 5-10 mA collector current used by cheap DMMs, Ebay component testers or even expensive small test instruments. They do not measure power transistors properly or anywhere near the current of interest.
Measuring the small capacitance from collector to base (Ccb) carefully, is a pretty good way to compare parts and even tell how sluggish the chip may be, if the capacitance is perhaps hundreds of pF rather than <100pF, for example.
It's typical advice to crack transistors open but:
1. how to crack them open without reducing them to an unrecognizable mess?
2. What constitutes a genuine part when you have only fakes for reference?
3. There are existing threads with pics posted of genuine and fakes already. Search easily here by the same thread title, or the parts forum where this thread belongs or on Google.
Remember guys, 2SC5200/A1943 are obsolete now. The Toshiba replacements, TTC5200/A1943 or 2SC5200N/A1943N, have dies that are about 30% smaller than original. Results are easily confused without better advice and experience with what the typical fakes are like now, so they can compared with accuracy. Otherwise, this could be wasting the OP's time.
Measuring the small capacitance from collector to base (Ccb) carefully, is a pretty good way to compare parts and even tell how sluggish the chip may be, if the capacitance is perhaps hundreds of pF rather than <100pF, for example.
It's typical advice to crack transistors open but:
1. how to crack them open without reducing them to an unrecognizable mess?
2. What constitutes a genuine part when you have only fakes for reference?
3. There are existing threads with pics posted of genuine and fakes already. Search easily here by the same thread title, or the parts forum where this thread belongs or on Google.
Remember guys, 2SC5200/A1943 are obsolete now. The Toshiba replacements, TTC5200/A1943 or 2SC5200N/A1943N, have dies that are about 30% smaller than original. Results are easily confused without better advice and experience with what the typical fakes are like now, so they can compared with accuracy. Otherwise, this could be wasting the OP's time.
I was comparing sound quality, not durability, and more importantly, between parts sold in my country.
Before the monetary crisis in 1998, there was no serious counterfeit issue. After the crisis, the MT-200 (and even K135/J50) are not as good sounding as it used to be. I believe in my ears, that's why I'm sure they are "fake" or "off-spec". They are officially sold, but at lower price.
Also in class AB, not only class-H. Near my house (walking distance) there are 2 electronic shops building amps for PA use. One shop has class-H the other shop uses just class-B (as their "best" demoed amplifiers). And they use local C5200 which they are selling (also the MT-200 Sankens). Originally I didn't believe the local 5200 can withstand high voltage, but those examples were the proof.
And every such practitioners (here, local practitioners with local parts) are agree that if you want sound you go with the Toshiba but if you want security (PA use) go with the Sankens. With local parts, it is so obvious.
In the Netherlands it really ain't a prob to get the real once and and like sjoerd said, they're kind of hard to read. But why don't you buy some and check the datasheet for max values and go 20% below that.... I assume the fake one will die. It's better then installing him and he then gives up. I also can send you a couples, since i bought them for making an amp, but they were a bit too positive thinking.
You can also find them often at stores where they sell dj/studio/pa amps, since they are used on a wide scale in the pa.amps.
Greets richard
You can also find them often at stores where they sell dj/studio/pa amps, since they are used on a wide scale in the pa.amps.
Greets richard
I think it is an issue with the Mega328.
In the 128 to 237 range of hFE is got just four different hFE values.
It seems that there is some bit approximation going on in the digital parts of the circuit and when converted to a read out gives only 4 values. These were 128, 151, 185 & 237.
I guess they were just 4bits of difference in a digital measurement.
For the low range of hFE , I got many more different hFE values:73, 75, 77, 78, 79, 80, 87, 89, 98, 111 & 112. Again guessing it had more current to work with and thus many more bits of measurement.
The big issue is no overlap between NPN and PNP. I used two 128 with 111 & 112 to make a 2pair output stage.
I had the same problem with Onsemi MJE15034/5
I will re-test the 1943/5200 using my old and slow analogue methods.
The Mega 328 seemed to have measured To92 OK. But there were a surprising number of devices with apparently identical hFE and Vf (Vbe?)
Not necessarily.
I have posted repeatedly about "batching" using a DMM hFE facility and that "matching" MUST be done at operating current. DMM readings are usually useless.
Output devices are biased to significant currents that far exceed test current values used in a machine designed to test ALL devices.
In some builds I have batched at very low currents (whatever the machine uses) and then properly matched at 100mA or 500mA or whatever the amplifier used as an operating bias current.
I have FETs for an F5X that are matched for point values of Vgs AND for slope. PMA organised the testing for the group buy. Worth reading about the efforts the Team put in to get good components for our builds.
If you look at complementary pairs of devices, both low power and medium power and high power you will mostly find that the hFE vs Ic do not match between complements at all usable currents.
very often one of the complements changes hFE to a much lower value as current goes down. The very low test current could well be highlighting the DIFFERENCE between the roll off regions of the hFE vs Ic. The hFE in the operating current range may turn out to be much closer.
It is worth learning how to read a datasheet.
You are completely misguided.
Look at the datasheets.
SOA for the 1943/5200 is terrible at Vce above 60Vdc.
Any high power ClassAB is wasting resources using these devices.
BUT
ClassG or H is quite different.
The devices only see a low Vce for much of the time and the high Vce is applied for very short durations. Be careful doing full power testing. One could easily exceed the short period durations and massively exceed the temperature de-rated SOAR in a few seconds at 1000W into 4r0.
ALL devices can pass much higher currents for very short periods compared to DC and 100ms specifications.
ClassG/H uses this short period transient capability. 1943/5200 work well if the Designer recognises the limitations and DESIGNs for these capabilities.
Last edited:
A few comments on measuring Hfe of power devices using that ATMEGA toy.
I recently (a) bought one of those atmega transistor testers off EBAY, and (b) bought a stack of power devices, in my case MJL21193/4 and MJL4301/4281.
The ATMEGA transistor tester seems to mostly work, but I remain unconvinced at its accuracy. It certainly is not giving repeatable results unless I turn it off between measurements.
Also, the Hfe of power devices when measured by simple transistor testers is misleading. The initial results I got for the power devices had Hfe values all over the shop.
Building a test jig and measuring Hfe at a decent current (well from low current up to about 6A) gave much more sensible results, and certainly much more like the Hfe values you need in designing an amplifier.
--> comment. What would make you want to match Hfe of output devices? Provided the Hfe is not low, I wonder what it really achieves for you. (matching input devices Hfe / Vbe does make sense).
--> PS AndrewT - agree 100%
I recently (a) bought one of those atmega transistor testers off EBAY, and (b) bought a stack of power devices, in my case MJL21193/4 and MJL4301/4281.
The ATMEGA transistor tester seems to mostly work, but I remain unconvinced at its accuracy. It certainly is not giving repeatable results unless I turn it off between measurements.
Also, the Hfe of power devices when measured by simple transistor testers is misleading. The initial results I got for the power devices had Hfe values all over the shop.
Building a test jig and measuring Hfe at a decent current (well from low current up to about 6A) gave much more sensible results, and certainly much more like the Hfe values you need in designing an amplifier.
--> comment. What would make you want to match Hfe of output devices? Provided the Hfe is not low, I wonder what it really achieves for you. (matching input devices Hfe / Vbe does make sense).
--> PS AndrewT - agree 100%
Last edited:
Notice the pads for pressing on sot23 and sot223 devices.
I checked a few sot23 mosFETs the other day to ensure the MEGA328 was able to do what they described.
The major failing so far is inductance. It displays rubbish values for DCR.
- Status
- Not open for further replies.