Can anyone help me with a J-fet problem? I need to replace some marked K147 and J72. I assume these are 2SK147 and 2SJ72, but I not been able to locat them in South Africa. These are located in the output stage of my DAC. I belive 2SK170 and 2SJ74 might do, and I have located some 2SK170's locally, but they can't tell me what make they are, nor can they provide me with matched pairs, nor can they supply 2SJ74's. Can anyone suggest other high quality pairs that I might be able to use? I have no idea what parameters I should be looking for.
http://www.borbelyaudio.com/audiophile_components.asp
http://www.reichelt.de/
Normally there are some extra letters (GR/BL/V) that tells something about the Idss, which can vary a lot.
http://www.reichelt.de/
Normally there are some extra letters (GR/BL/V) that tells something about the Idss, which can vary a lot.
Those parts numbers indeed have a "2S" prefix. They were made by Toshiba, but have been discontinued for over 10 years now. The only way to find them is through old stock from a distributor. When you do find them, they are expensive.
The closest equivalents are the 2SJ74 and the 2SK170. Paralleling two of those will be pretty much the same as your original parts. The problem is that you have to match the paralleled parts, which can be impractical for the home constructor.
I am surprised that they have failed. Perhaps the circuit design is poor and the devices are too highly stressed. We use the Toshiba JFETs in many of our designs. The only time I have seen any of the single JFETs fail is when a customer plugged in a high output CD player (10 volts RMS!) into the MC phono input. Even then, the input transistors weren't damaged, but instead the cascodes.
The closest equivalents are the 2SJ74 and the 2SK170. Paralleling two of those will be pretty much the same as your original parts. The problem is that you have to match the paralleled parts, which can be impractical for the home constructor.
I am surprised that they have failed. Perhaps the circuit design is poor and the devices are too highly stressed. We use the Toshiba JFETs in many of our designs. The only time I have seen any of the single JFETs fail is when a customer plugged in a high output CD player (10 volts RMS!) into the MC phono input. Even then, the input transistors weren't damaged, but instead the cascodes.
Thanx for the info. Something has failed. I was not sure what. These were the closest devices to the output, so I was going to try replacing them to see if that sorted out the problem - IF I could find cheap replacements easily. Since I can't, I shall have to look elsewhere for the fault. Since you are surprised these might have failed, this confirms that they probably haven't.
Any ideas what might go first when given an unexpected current back into the analog outputs? I don't see any burnt components, but I am measuring 0,6 Ohms between output and ground. That can't be right, can it?
Any ideas what might go first when given an unexpected current back into the analog outputs? I don't see any burnt components, but I am measuring 0,6 Ohms between output and ground. That can't be right, can it?
What about 2SK389/2SJ109? They're very similar to 2SK170/2SJ74, but being dual parts they can be paralleled with no problem.Charles Hansen said:
hihopes said:
Hmm. I'm not sure what you mean by "current back into the analog outputs", but this could definitely cause problems. Assuming that the power is disconnected, measuring 0.6 ohms to ground is not good unless there is a muting relay that shorts the output to ground.
There are two ways to measure the FETs -- in circuit and out of circuit. If you remove them, they should measure like a diode between gate and either the drain or the source. From drain to source should be a few thousand ohms give or take an order of magnitude or so.
In circuit, I like to measure with power applied. The source should be within 100 mV of the gates for these devices. The drain voltage will depend on the circuit, but no matter what you should have something like 5 to 15 volts from the drain to the source.
Good luck!
Mr Evil said:
On paper, yes. However, there are a few problems in actuality:
1) These dual parts were discontinued last year. It may be difficult to find them (although probably not any more difficult than the original parts).
2) The packaging would make it difficult to parallel the pins. It would be easier to "piggyback" a pair of single FETs.
3) The substrate is separate on the dual FETs. You have to figure out how to handle the substrate pin. (The data sheet is *not* the best resource here.)
4) I have had reliability problems with these parts. They seem to be prone to damage from static discharges.
Thanks for the help, and thanks for wishing me luck. I now place myself in the hands of the hi-fi gods.
Hi Charles,
what do you think ist the best connection when using these parts in a LTP? Connecting the substate together with the sources?
You will have to measure the connection from the substrate pin to the other pins with the "diode" mode of a multi-meter. A parasitic diode exists between some connections, which should not be forward biased. Once you have determined which connnections are allowable, you will need to do some listening tests (like everything else in audio!) to see what connection sounds best.
Thank you Charles for this hint.
I appreciate very much that you share this kind of information with us.
Tino
I appreciate very much that you share this kind of information with us.
Tino
Hi Charles,
I thought some times before of shielding the FET itself by the substrate, not only left this pin floating as it it recommended in the Toshiba datasheet. In some circuits it might be possible to ground the substate (inverting op-amp for example). I remember the old RCA CA3046 datasheets where the substrate should connected to the most negative potential.
I thought some times before of shielding the FET itself by the substrate, not only left this pin floating as it it recommended in the Toshiba datasheet. In some circuits it might be possible to ground the substate (inverting op-amp for example). I remember the old RCA CA3046 datasheets where the substrate should connected to the most negative potential.
- Status
- Not open for further replies.