FET Questions=Martin?

[Charlie Wilson]

Well a long time ago I set my FET as Martin suggested and was happy that I could simply get it to work. Now I want to fine tune it and possibly make it footswitchable but I have a couple questions. I ended up with a 24k and 2k2 bias resistor with a NTE 452. You suggested that the bypass 4.7uf cap would be increased to 22uf to duplicate the frequency response of Dumble's set up but with a different FET. Would that still apply to the NTE 452 as well? Also my board has a 47uf/50v filter cap but I see that most the Dumble boards have a 100uf/50v. Finally, Dumble seemed to like to use a 1/2w resistor for the 150k. I did the math and that would put it right at the wattage limit of the resistor. Bad idea?(I think I know the answer to that one )
CW

[martin manning]

Well a long time ago I set my FET as Martin suggested and was happy that I could simply get it to work. Now I want to fine tune it and possibly make it footswitchable but I have a couple questions. I ended up with a 24k and 2k2 bias resistor with a NTE 452. You suggested that the bypass 4.7uf cap would be increased to 22uf to duplicate the frequency response of Dumble's set up but with a different FET. Would that still apply to the NTE 452 as well?

It depends on how you want to shape the boost frequency response at the low end, same as for a triode gain stage. The exact FET type won't matter much.

Second Gen layout shows 2N3823 with 4u7//1k6 on the source, all the others show NTE452 with 4u7//8k2. On the icracer 124 schematic, Rs is 8k8. In practice NTE452 seems to bias with around 2k, as does 2n4416A, for which NTE452 is a substitute.

My thought was if the 8k//4u7 was correct on the later originals, and the required resistance to bias was reduced to about 2k, then the cap would need to be 4x as big. Based on recent experience 8k seems like an outlier, so it's hard to know what the original intent was. A 4u7 will result in -3dB point around 75Hz, and will go inversely with the cap value. With 22u, you'll get a full range boost with -3dB at 16Hz. Seems like the Second Gen FET would have been a very trebly boost.

The other major contributor to the low frequency roll off is the drain coupling cap (4u7) and the load (10k), but that is out of the way (around 10Hz).

Also my board has a 47uf/50v filter cap but I see that most the Dumble boards have a 100uf/50v.

B+5 is already very quiet, so I don't think there is a need for 100u, even 22u would probably be fine.

Finally, Dumble seemed to like to use a 1/2w resistor for the 150k. I did the math and that would put it right at the wattage limit of the resistor. Bad idea? (I think I know the answer to that one :D )

I agree with you, it should be a 1W at least. The #124 photos show a large CF resistor. Tony's layouts show a large resistor without a power rating. On icracer's 124 schematic it's marked 2W.

[talbany]

1W CC

Tony

[Charlie Wilson]

Thanks Martin. Yeah Tony I think that is what it should be, but in 124 and Ry Cooder and all of the other Fets I have seen he uses 1/2w.
CW

[Charlie Wilson]

Martin I want to ask you about this;
My thought was if the 8k//4u7 was correct on the later originals, and the required resistance to bias was reduced to about 2k, then the cap would need to be 4x as big. Based on recent experience 8k seems like an outlier, so it's hard to know what the original intent was. A 4u7 will result in -3dB point around 75Hz, and will go inversely with the cap value. With 22u, you'll get a full range boost with -3dB at 16Hz. Seems like the Second Gen FET would have been a very trebly boost.
I set my divider resistor at 24k to get the 20v you suggested and the FET works fine no problems. However, if I did use the 150k/8k2 divider and used a larger(than 2k) bias resistor would I not still be able to get the 10v on top of the drain just with less voltage drop across the 10k? The only reason I mention this is that I had a friend measure the voltage at the bottom of the 150k resistor on a real Dumble yesterday and it was a tad over 10v. He wasn't able to get the probe on the other side of 10k to see how much voltage was droppped. He also was able to measure the voltage on the top of the bias resistor and it was 2.46v which is double the voltage I get on mine.
CW

[martin manning]

Well, knowing that there is 10V on the drain is a good data point. Can you get the value of the source resistor on you friends amp?

[talbany]

but in 124 and Ry Cooder and all of the other Fets I have seen he uses 1/2w.

Yeah I know! I was just recommending a 1 W CC. Thanks for the shots

Tony

[Charlie Wilson]

Well, knowing that there is 10V on the drain is a good data point. Can you get the value of the source resistor on you friends amp?

Yes. I’ll let you know when I get it.
CW

[martin manning]

Well, knowing that there is 10V on the drain is a good data point. Can you get the value of the source resistor on you friends amp?

Yes. I’ll let you know when I get it.

Great. Anything else you can find out would help too, like the source cap and drain resistor values, and the FET type.

[Charlie Wilson]

Sounds good. I'll get as much as I can. It looks very similar to the first photo I posted with the NTEs.
CW

[Charlie Wilson]

Martin,
FET NTE 452
Drain 10k ( top 10.48v/ bottom 3.82v)
Source 3.3k ( 2.182v)
Caps were both the same sized tantalum. So I am assuming 4.7uf
B+ on top of the 150k was 302v(150k/8.2k divider)
All NTE resistors
The FET sounded really good. Strong but still transparent.
CW

[martin manning]

Ok, good stuff there. We can see it is running a low supply voltage, 10.5 Vdd, and the chosen bias point drops more than half of the supply voltage across the drain resistor. That says it won't have the clean headroom that a higher Vdd would bring, it's not well centered, and it will produce a lot of its own distortion. I'm not sure where the idea that the Vdd should be 18-20V came from (it's from years ago), but the NTE FET used has Vgs of 30V (same as 2N4416A), and it can easily operate there. Higher Vdd and a centered bias point maximizes clean headroom and transparency. Many people including myself have set the FET up at higher Vdd (18-20V), and liked the clean boost and its ability to overdrive the first triode stage.

There is another set of voltages available in the SSS 002 hand-drawn schematic. The supply divider shown there is 220k/8k2, giving just 7.4V Vdd. Vd is 3.77, which is very close to half of Vdd. The drain resistor is 10k, and the drain coupling and source resistor bypass caps are 5uF. The source resistor is 2k7, and the FET is marked F1175. So, there's a case with even lower Vdd, but it's center biased.

The constant features seen on all generations seem to be the 8k2 lower divider leg, a 10k drain resistor, 10k level pot, and 4u7/5u caps. SSS 002 is the only one I've seen with the 220k upper divider leg. It seems that HAD's technique was to fix all of those values and adjust only the source resistor value to set the bias point, letting both the Vdd and the low frequency response go where they may. The bias point could have been set by eye (scope), or by ear.

PS Maybe it would be a good idea to move this discussion under the FET bias thread in the files section?

[Charlie Wilson]

So if one were to want to try the Dumble set up with the 150k/8.2k divider, would a good starting point be find a value for the bias resistor that gives you half the voltage on the bottom of the drain resistor then is on the top at whatever the divider voltage ends up being? Not questioning that your setup is ideal, just trying to get a feel for what Dumble might have been going for. It looks like on the Ry Cooder FET the bias resistor is 1.3k. I realize after writing this that your battery method does what I suggest already just with a target voltage of 20v.
CW

[martin manning]

So if one were to want to try the Dumble set up with the 150k/8.2k divider, would a good starting point be find a value for the bias resistor that gives you half the voltage on the bottom of the drain resistor then is on the top at whatever the divider voltage ends up being?

Yes, that will work fine as long as B+5 is going to be around 300V. Choose the value for the lower resistor in the divider (8k2 for low Vd, or say 18-20k if you want high Vd), temporarily connect a pot (10k) for the source resistor, dial in the drain resistor voltage ratio, then measure and replace the pot with a fixed resistor.

[Charlie Wilson]

A little off topic but Martin would you consider the 220k/8k2 divider a good way to get the preamp voltages up a bit. My amp needs it a bit and it would be great to not have to tear my filter board a part. I guess I would have to figure out what an equivalent divider would be to the 8k2.
CW