| GregGC |
I was looking for a way to decrease the input cap and not use a NFB cap. So that's what I came up with (didn't see similar sch. in the forum so decided to share it). You can acheave easily more than 100k input impedance and use 1uF (-3dB@1.5Hz) or less, input cap of high quality and not spend too much $$$.
The sch will be good for application where a higher Zin is needed (maybe tube preamps).
The gain of the amp is set by R7/R6 (22 in this case). For higher gain increase R7.
If you need higher impedance (I wouldn't go higher than 470k) increase R3 and R4 at the same time to the needed value.
To get lower output DC offset you need to match R3 and R4 as close as possible (1% or even better is preferable).
Comments appreciated.
/Greg |
|
|
| matjans |
wouldn't HF noise become a problem?
on my pre at least (i've built a couple a la carlosfm) the high-impedance versions tend to pick up lots of radio where the lower (<100K) didnt...
Using very well shielded interconnects helps out a lot, but still. |
|
|
| GregGC |
matjans,
It would be more suseptable to RF noise (because of the higher Zin), but if you need high impedance you take the necessary precautions too. It should be well shielded anyway. If it's in a wooden box (as I've seen way to often here) it's not good even for 20k Zin anyway and also I always use the input RC network(470/1nF) to kill some of those RF nasties. If the GC is a part of an integrated amp, then everythig is inside a unified metal box anyway. I think a lot of people, that build the GC don't pay enough attention to the shielding an the RF nois to begin with, which deteriorate the sound quality.
/Greg |
|
|
| GregGC |
Another way to decrease the RFI by decreasing the input impedance is to use a lower value pot (25k for example) infront of the 1uF cap. That way you still have 25k Zin and the input cap is still 1uF because it sees 100k impedance. The best of both worlds.
/Greg |
|
|
| matjans |
true greg, very true. Although the difference in price for the input cap is not *that* big... In the Netherlands that is. A 4,7uF BG-N is about EUR 2,65 and a 1uF BG-N is EUR 2,10. Not that much of a difference and considering the GC-slogan is 'less is more' i'd just go for the 4,7uF version. Saves you one resistor. ;)
But still, with valve preamps this might be a really good idea. Does anybody have experience with high/low impedance inputs on amps in combination with a valve preamp? I sure don;t.
anyways, i had a lot of troubles getting rid of rfi but i managed. my clone is dead silent at the moment.
Only thing is I still *have* to use shielded interconnects although my input impedance is (only) 22K; (i use a sort-of diy xlr interconnects; interlink shielding is all connected to the clone's case, which in turn is connected to ground on the PSU). El cheapo $1 interconnects cause rfi.
too bad. On the other hand, who wants to use those interlinks anyway? I'm more than happy with my (grand total of 25EURO;S worth ) diy silver/teflon/copper shielding interlinks.
M. |
|
|
| GregGC |
| quote: | Originally posted by matjans
true greg, very true. Although the difference in price for the input cap is not *that* big... In the Netherlands that is. A 4,7uF BG-N is about EUR 2,65 and a 1uF BG-N is EUR 2,10. Not that much of a difference and considering the GC-slogan is 'less is more' i'd just go for the 4,7uF version. Saves you one resistor. ;)
But still, with valve preamps this might be a really good idea. Does anybody have experience with high/low impedance inputs on amps in combination with a valve preamp? I sure don;t.
anyways, i had a lot of troubles getting rid of rfi but i managed. my clone is dead silent at the moment.
Only thing is I still *have* to use shielded interconnects although my input impedance is (only) 22K; (i use a sort-of diy xlr interconnects; interlink shielding is all connected to the clone's case, which in turn is connected to ground on the PSU). El cheapo $1 interconnects cause rfi.
too bad. On the other hand, who wants to use those interlinks anyway? I'm more than happy with my (grand total of 25EURO;S worth ) diy silver/teflon/copper shielding interlinks.
M. |
Mat,
You're right. In my GC I don't use any caps at all, but I'd like to try the higher impedance AC in version.
Aren't those BG-N caps electrolytic Non-Polar type?
If using film type the difference in size between 1 and 4.7uF is huge though.
I've always used shielded interconnects (nothing fancy) no matter what the Zin. Never had problems with the RFI.
/Greg |
|
|
| theChris |
i've made a high input impedance IGC. cuttoff is somthing like 3hz. 1uF cap, 100k input impedance. unfortunately the dc offset was 50mV (70mV for bridged).
the only issues with your circuit is that the DC gain is still 10+, so if the amp nominally would have 1mV of offset, it again has 10mV. you do address the input bias offset, which i agree is good practice, but for large value of bias resistors, the DC offset becomes a function of the internal voltage offset, Vos.
i think BG-N is for "normal" black gate caps, not "non-polarized". though i could be wrong. |
|
|
| matjans |
| you are chris ;) I don't know what the N stands for but they are non-polar types. |
|
|
| GregGC |
| quote: | Originally posted by theChris
i've made a high input impedance IGC. cuttoff is somthing like 3hz. 1uF cap, 100k input impedance. unfortunately the dc offset was 50mV (70mV for bridged).
the only issues with your circuit is that the DC gain is still 10+, so if the amp nominally would have 1mV of offset, it again has 10mV. you do address the input bias offset, which i agree is good practice, but for large value of bias resistors, the DC offset becomes a function of the internal voltage offset, Vos.
i think BG-N is for "normal" black gate caps, not "non-polarized". though i could be wrong. |
Chris,
Your right about the DC gain. That's why I think the matching of the two 100k res is quite important. If they are within 1% or less it should be fine. What tolerance res. did you use? About the NFB Res., are they 100/2.2k or different values (because the 100 adds to the 100k)?
I think those BG-N are toos mall to be anything else but Non-Pol electrolitycs.
/Greg |
|
|
| theChris |
the circuit i used was a t network. 100 k in, and i think 20k,30k and 300 ohms. not sure though, they were value i had around. 1% matching.
and the N was nonpolarized. the "standard" line was what i was thinking of. |
|
|
| GregGC |
Chris,
I look at it that way: If you look at the NIGC the Res from +in to GND is 20k and the -in to GND is 1k. so the dif is 20k and that creates around 30-50mV DC at the output. In the Hi-Impedance GC if the 100k res to GND ar matched to less than 1k (1%) should be less that 10 mV output DC offset. That's what I get simulating the circuit also (less than a 1mv. and 15mV if the there is a 2k dif between the two resistors).
/Greg |
|
|
|
