Post by thread:CCFL inverter transistor replacement

> ... from NEC I believe, ie 2SD882 on your diagram. > ... I replaced it with a BD437, and the inverter worked , but the > transistor got pretty hot > (certainly hotter than the original). ... > Here's the schematic, with the original transistor: > http://marcansoft.com/subidos/ccfl_inverter.png It's not clear why it would be getting hotter as it's not obvious that the BD437 is inferior in any important ways. Current gain seems the most likely culprit. The claimed current gain is down somewhat - say 40/?70? min/typical at 2 amps compared to 60/160 at 1A for the original. But Vsat is superior. Vce is slightly lower but should not matter. It may be that they select on test the 2SD882 or that they get ones which have gains in the higher spec sheet range - its not uncommon for devices to be improved with time and move towards the higher end of a spec range. If current gain is the issue then selecting from a batch of BD437s with a beta-meta (sounds catchy) may be useful. Yu MIGHT be able to get a logic level FET working there - you'd probably need a from gate (was base) to ground. This would solve any current gain issues but may have other problems. I'm not sure what power/current level you are working at but an Rdson of 0.1 ohm or less would be OK and quite a bit more may be OK. Vdrive transformer.3_5 is unknown of course but probably enough to give 1 to 2 mA in the 5K6 so may be in the 5-10 volt range, which would make a FET viable. Something like an XXX3055L low voltage logic level-cheapo N Channel MOSFET would be an OK place to start. But measure Vt.3_5 and see what sort of drive voltage is available. Even a non-logic FET may work. Russell

On 7/9/06, Russell McMahon <spam_OUTapptechTakeThisOuTparadise.net.nz> wrote: > > ... from NEC I believe, > > ie 2SD882 on your diagram. > > > ... I replaced it with a BD437, and the inverter worked , but the > > transistor got pretty hot > > (certainly hotter than the original). ... > > > > Here's the schematic, with the original transistor: > > http://marcansoft.com/subidos/ccfl_inverter.png > > > > It's not clear why it would be getting hotter as it's not obvious that > the BD437 is inferior in any important ways. Current gain seems the > most likely culprit. 2SD882 it's a real switching transistor with a 90Mhz bandwidth while BD437 it's a 3Mhz one. If you need to know more, then visit a romanian site made by a guy who knows what he want: http://www.datasheetcatalog.net/cgi-bin/helo.pl?field=Nume&type=C&text=SD882&producedby=&action=Search http://www.datasheetcatalog.net/cgi-bin/helo.pl?text=BD437&action=Search greetings, Vasile

Russell McMahon wrote: >> ... from NEC I believe, > > ie 2SD882 on your diagram. Forgot to fix the message after noticing I had the reference marked on the diagram from back then. It's a 2SD882 of course (took me forever to find, since all that was written on the front was D882...) {Quote hidden}> >> ... I replaced it with a BD437, and the inverter worked , but the >> transistor got pretty hot >> (certainly hotter than the original). ... > > >> Here's the schematic, with the original transistor: >> marcansoft.com/subidos/ccfl_inverter.png > > It's not clear why it would be getting hotter as it's not obvious that > the BD437 is inferior in any important ways. Current gain seems the > most likely culprit. > > The claimed current gain is down somewhat - say 40/?70? min/typical > at 2 amps compared to 60/160 at 1A for the original. But Vsat is > superior. Vce is slightly lower but should not matter. It may be that > they select on test the 2SD882 or that they get ones which have gains > in the higher spec sheet range - its not uncommon for devices to be > improved with time and move towards the higher end of a spec range. If > current gain is the issue then selecting from a batch of BD437s with a > beta-meta (sounds catchy) may be useful. Would you recommend another transistor as a replacement? I have access to a moderately-sized electronics supplier here in spain, so they should have most common stuff, and they get free shipping from digi-key so I can get anything that they carry (with a delay of a week or so, which is a pain but I can handle in this case) {Quote hidden}> > Yu MIGHT be able to get a logic level FET working there - you'd > probably need a from gate (was base) to ground. This would solve any > current gain issues but may have other problems. I'm not sure what > power/current level you are working at but an Rdson of 0.1 ohm or less > would be OK and quite a bit more may be OK. > > Vdrive transformer.3_5 is unknown of course but probably enough to > give 1 to 2 mA in the 5K6 so may be in the 5-10 volt range, which > would make a FET viable. Something like an XXX3055L low voltage logic > level-cheapo N Channel MOSFET would be an OK place to start. But > measure Vt.3_5 and see what sort of drive voltage is available. Even a > non-logic FET may work. Could I set up a non-logic FET together with some booster transistors to bring the level up to what the FET handles? I have a bunch of IRF510's here, and lots of random low-power transistors. Otherwise I'll try with some logic-level FETs. I was also wondering... Could the problem be current being dissipated into the transistor some awkward way, that the original tolerated (i.e. shorted) better? Like, something a diode could fix. I don't know exactly how these oscillators work, but could I have some back-EMF ending up where I don't want it? -- Hector Martin (hectorKILLspammarcansoft.com) Public Key: http://www.marcansoft.com/hector.asc

At 12:29 PM 7/9/2006 +0200, you wrote: >Vasile Surducan wrote: > > > > 2SD882 it's a real switching transistor with a 90Mhz bandwidth while > BD437 it's > > a 3Mhz one. > >I thought 3Mhz would be plenty... Unfortunately I don't have a decent >frequency counter around, but I'd hope it isn't trying to drive the CCFL >at 10Mhz (how much RFI can that produce? This is supposed to be used >inside PCs...) No, a 3MHz ft is not plenty-- it's barely good enough for analog low-end audio, like the ancient 2N3055. ft is *unity gain* cutoff frequency, so a 3MHz ft transistor is pretty much useless at 3Mhz, and would have low gain even with a 300kHz sine wave. The blocking oscillator is probably operating at 50-100kHz. Change the transistor to an appropriate type and it will work well again. The D882 is USD 0.74 from Digikey in quantity 1. >Best regards, Spehro Pefhany --"it's the network..." "The Journey is the reward" .....speffKILLspam.....interlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com ->>Test equipment, parts OLED displys http://search.ebay.com/_W0QQsassZspeff

On Sun, 9 Jul 2006, Russell McMahon wrote: > The claimed current gain is down somewhat - say 40/?70? min/typical at 2 > amps compared to 60/160 at 1A for the original. But Vsat is superior. Vce is > slightly lower but should not matter. It may be that they select on test the > 2SD882 or that they get ones which have gains in the higher spec sheet range > - its not uncommon for devices to be improved with time and move towards the > higher end of a spec range. If current gain is the issue then selecting from > a batch of BD437s with a beta-meta (sounds catchy) may be useful. Ime European replacements for Japanese transistors almost always suck. I don't know exactly why this is so but it is so. Peter

You could try BD139 or MJE180. On 9 Jul 2006 at 20:34, Peter wrote: {Quote hidden}> > On Sun, 9 Jul 2006, Russell McMahon wrote: > > > The claimed current gain is down somewhat - say 40/?70? min/typical at 2 > > amps compared to 60/160 at 1A for the original. But Vsat is superior. Vce is > > slightly lower but should not matter. It may be that they select on test the > > 2SD882 or that they get ones which have gains in the higher spec sheet range > > - its not uncommon for devices to be improved with time and move towards the > > higher end of a spec range. If current gain is the issue then selecting from > > a batch of BD437s with a beta-meta (sounds catchy) may be useful. > > Ime European replacements for Japanese transistors almost always suck. I > don't know exactly why this is so but it is so. > > Peter > --

On 7/10/06, Xiaofan Chen <KILLspamxiaofancKILLspamgmail.com> wrote: > On 7/10/06, Vasile Surducan <RemoveMEpiclist9TakeThisOuTgmail.com> wrote: > > On 7/9/06, Peter <spamBeGoneplpspamBeGoneactcom.co.il> wrote: > > > Ime European replacements for Japanese transistors almost always suck. I > > > don't know exactly why this is so but it is so. > > > > Let me correct you: not almost always but ALWAYS. > > > > Vasile > > What is the reason? Is it because the Japanese vendors' datasheet > almost always sucks unless it is written outside of Japan? Data sheet may sucks as long the transistors are the best. One example are the HV transistors used in 21'' monitors. The other one is the comparison between any medium power japanese transistor versus worldwide made transistor, japanese have 5x to 10x bigger cut off frequency, which means the japanese technology is cleaner than the others. Replacing most of the japanese transistors are a real pain for an european/american user. Vasile

On 7/10/06, Vasile Surducan <TakeThisOuTpiclist9EraseMEspam_OUTgmail.com> wrote: > > What is the reason? Is it because the Japanese vendors' datasheet > > almost always sucks unless it is written outside of Japan? > > Data sheet may sucks as long the transistors are the best. > One example are the HV transistors used in 21'' monitors. > The other one is the comparison between any medium power > japanese transistor versus worldwide made transistor, japanese have 5x to 10x > bigger cut off frequency, which means the japanese technology is > cleaner than the others. > Replacing most of the japanese transistors are a real pain for an > european/american user. > In this case, an easy solution is to avoid using the Japanese transistors and use generic European/American transistors if possible.

Spehro Pefhany wrote: > At 12:29 PM 7/9/2006 +0200, you wrote: > >> Vasile Surducan wrote: >>> 2SD882 it's a real switching transistor with a 90Mhz bandwidth while >> BD437 it's >>> a 3Mhz one. >> I thought 3Mhz would be plenty... Unfortunately I don't have a decent >> frequency counter around, but I'd hope it isn't trying to drive the CCFL >> at 10Mhz (how much RFI can that produce? This is supposed to be used >> inside PCs...) > > No, a 3MHz ft is not plenty-- it's barely good enough for analog low-end > audio, like the ancient 2N3055. > > ft is *unity gain* cutoff frequency, so a 3MHz ft transistor is pretty much > useless at 3Mhz, and would have low gain even with a 300kHz sine wave. The > blocking oscillator is probably operating at 50-100kHz. Change the transistor > to an appropriate type and it will work well again. The D882 is USD 0.74 > from Digikey in quantity 1. That would explain things. Mental note: do not assume that parameters mean what you think they mean. I'm guesstimating the inverter runs at about 30-40khz, from the measurements my (crappy) multimeter-integrated frequency counter gives me, and the fact that at a lowered input voltage I can actually hear the transformer whine (that would be 18khz or so). -- Hector Martin (RemoveMEhectorTakeThisOuTmarcansoft.com) Public Key: http://www.marcansoft.com/hector.asc