Pass JFET Power Amplifier

Status
Not open for further replies.
Jump to Latest
#122
You were apparently trying to tell us why the 2N6550 was a better choice. The main advantage you were claiming is that you could use fewer of them.

But using the 2N6550 would create a lower transconductance (yielding higher output impedance and higher distortion), as well as have a higher input capacitance. You will save some time stuffing the board. But I bet the Toshibas cost less than half the American parts, so you wouldn't save any money. Not a good trade-off in my book. But it's your amp. Go for it.
 
#123
Charles Hansen said:
You were apparently trying to tell us why the 2N6550 was a better choice. The main advantage you were claiming is that you could use fewer of them.
Click to expand...

I wasn't trying to tell anybody anything. I was only saying that the differences weren't that great and that FOR ME I would prefer to use fewer devices if possible. I wasn't trying to tell anyone else what choices THEY should make.

But using the 2N6550 would create a lower transconductance (yielding higher output impedance and higher distortion), as well as have a higher input capacitance.
Click to expand...

How do you get lower transconductance? Transconductance doesn't add with devices in parallel does it? It does increase with current though, yes? And with less than half the devices, they'll be running at more than twice the current. So how are you getting lower transconductance?

Yeah, output impedance will be somewhat higher because there will be fewer devices in parallel. But how low does it need to be? If we want really low, why not parallel 500 or more of the 2SK170s? Why stop at 125?

And yeah, input capacitance will be a bit higher, but not by a huge amount.

You will save some time stuffing the board. But I bet the Toshibas cost less than half the American parts, so you wouldn't save any money.
Click to expand...

And that just brings me back again to the rather low IDSS of the 2SK170s. The 125 devices I mentioned was the ideal, based on each device running at 10mA. But of course unless all the devices are matched, some will be drawing a bit more and some will be drawing a bit less than 10mA. I'd want the devices I used to have an I[sub[DSS[/sub] somewhat greater than 10mA. Preferably I'd like something closer to 20mA.

Since the maximum IDSS of the 2SK170 is only 20mA, how many pieces do you think I would have to buy in order to get 125 pieces with an IDSS sufficiently above 10mA?

If I knew I wouldn't have to buy several hundred pieces to get a useable 125 then it'd be a bit more attractive to me. But as it stands I know I could buy 50 pieces of the 2N6550 and be pretty sure that all of them will have a more than sufficient IDSS.

se
 
#125
Charles Hansen said:
Have it your way, Steve. I'm trying to help and all you want to do is argue with me. Unfortunately, your arguments are based on false assumptions.
Click to expand...

I appreciate that you're trying to help. But you're starting to come across more as if you're dictating rather than offering advice. It almost seems that you're offended that I might choose something that you would not.

You say my arguments are based on false assumptions. Ok, what false assumptions are those? You say you're trying to help but then you say things like this and leave me guessing.

Go for it. By the way, the price for the 2N6550 from Crystalonics is only $18.75 each. Go for it.
Click to expand...

But from InterFET, they're only a couple dollars each. Apparently the Crystalonics devices are spec'ed for aerospace use or something. At least that's the best excuse I was able to get from one of their sales reps when I asked why their 2N6550s were so vastly more expensive than InterFET's.

se
 
#128
Oy . . .
Why Cgd is concern? I hardly understand it.
Acc to Nelson, one role of the cascode is to remove AC between Gate and Drain so to reduce distorsion in high frequencies. If so, I would not care the size of Cgd.

And . . .
I hardly understand where I put the importance of the transconductance.

😕

Regards
jH
 
#131
Steve Eddy said:
Transconductance doesn't add with devices in parallel does it?
Click to expand...


yes and no. Transconductance of a given device doesn't change when you parallel a bunch of them. However, the transconductance of the parralleled devices is essentially the sum of transconductance of those individual devices -> transconductance increases as you parrallel devices.

The same holds true for transistors (beta or hfe).
 
#134
Nelson Pass said:


The TA-7 is interesting in that regard, they use power JFETs
but not as gain devices.....

I suspect it worked very well nonetheless given the cascoding
of very fine bipolars.

😎
Click to expand...



I think it works pretty well. In engineering terms it is quite a complex beast, this TA-7. But it is well behaved, no oscillatory trends, never had a problem with it even on the test bench with funny loads and extensive high frequency tests. It is also the first amp ever I put wholesale in the dish washer, it came out as new and worked flawlessly 😀 .

Jan Didden
 

Attachments

  • tan7 bias.jpg
    tan7 bias.jpg
    91.4 KB · Views: 1,222
#137
jh6you said:
Oy . . .
Why Cgd is concern? I hardly understand it.
Acc to Nelson, one role of the cascode is to remove AC between Gate and Drain so to reduce distorsion in high frequencies. If so, I would not care the size of Cgd.

And . . .
I hardly understand where I put the importance of the transconductance.

😕

Regards
jH
Click to expand...


1 , they want to use jfet in follower for last output , if use cascode ,then need twice quantity of jfet .
2 the low gm also produce a relatively large Zout in a follower configuration. (original word in book <art of electronics> page 131
 
#138
Hi john-china,

Thank you for the info.
By the way, I thought that the discussion was to seek for the equivalent JFETs at the input of the Pass JFET Amp. Ayo . . . I used to go to school exam without thorough reading of books. My bad habit! 😀

Regards
jH
 
#139
JH6You,
i notice the problem with people tought to read in the other direction. (still no Lumanauw, btw)

The front part of Janneman's TA7 i would like to view too.
Same output fet-bipolar cascoding as Mr Thagard did with his design.
Those 473/1173's are amazing, 100 Meg and 3 amps, so amazing i snatched me some of the remaining original Toshiba's.

I wonder if NorGerman Borbely is monitoring this thread ? :clown:
 
Status
Not open for further replies.
Top Bottom