Yeah, Riken makes some really nice resistors.. All the rage in Japan supposedly? I got some from partsconnexion. They are precision carbon film, 1/2 watt, 1.0%, axial, with gold-plated copper leads. A bit more expensive than the Xicon though.dorkus said:
--Jordan
update: hah, Peter beat me to it (mentioning the riken's).
Actually, to prevent oscillation, and keep the amp (or any amp for that matter) stable at all times when using inverting mode, the gain must be >10 (or the spec of your amp).
If the input to your device is a pot, connected to a blocking cap, connected to an input resistor, the input impedance to the amplifer changes with movement of the pot from the maximum combined resistance of all components, to the resistance of just the imput resistor, when the pot is at maximum.
If the value of the maximum resistance, in concert with the feedback resistor provides your minimum gain of 10, you're fine. If it doesn't, the amp will be unstable at input values less than maximum where the gain of the amp is less than 10.
The remedy is to insert another resistor from the inverting input to ground, whose reisistance, combined in parallel with the maximum resistance of the input circuit, provides the amp with a gain of >10. The only drawback will be an increase in noise (probably undetectable) through the extra resistor, but this much more desireable than an oscillating amplifier. In all actuality the noise increase will probably not be either measureable or audible.
If the input to your device is a pot, connected to a blocking cap, connected to an input resistor, the input impedance to the amplifer changes with movement of the pot from the maximum combined resistance of all components, to the resistance of just the imput resistor, when the pot is at maximum.
If the value of the maximum resistance, in concert with the feedback resistor provides your minimum gain of 10, you're fine. If it doesn't, the amp will be unstable at input values less than maximum where the gain of the amp is less than 10.
The remedy is to insert another resistor from the inverting input to ground, whose reisistance, combined in parallel with the maximum resistance of the input circuit, provides the amp with a gain of >10. The only drawback will be an increase in noise (probably undetectable) through the extra resistor, but this much more desireable than an oscillating amplifier. In all actuality the noise increase will probably not be either measureable or audible.
dorkus said:
Maybe, but for the price I think they were the best deal around. Sonic Frontiers were using them in most of their equipment.
I once tried to replace Holcos with Caddock MK132 in I/V stage of my DAC, but quickly returned back to Holcos. The Caddocks sounded very boomy and unpleasant, while Holcos seemed to be more smooth. In this case Caddocks provided too much warmth for my taste and none of the vibrancy. I'm ordering large amount of Vishays S102 from the recent ad on the forum. Will see how they sound.
Rikkens seem to be kinda soft and warm, without metalic glare of the Holcos. But still no really consistant tests done. Although I have all those resistors on hand I somehow can't get myself to test them. You know what I mean?
Thanks for reply.
Here's another silly question. What if I remove the pot and connect the signal directly to the decoupling cap with all three remaining resistors of 47k? I tried this configuration with LM3875T (non-isolated package) and didn't have any oscillation (if the word "oscillation" means high frequency noize). What I'm trying to do is to set the input impedance of gainclone to ~47k because the output impedance of preamp is ~1k. Somewhere on the internet, I've seen that setting the input impedance of an amp to 50 times of output impedance of previous stage is good. BTW, with LM3875T with this configuration, the amp gets very hot. Is it because of low gain or what? I'm sorry for my poor knowledge.
Here's another silly question. What if I remove the pot and connect the signal directly to the decoupling cap with all three remaining resistors of 47k? I tried this configuration with LM3875T (non-isolated package) and didn't have any oscillation (if the word "oscillation" means high frequency noize). What I'm trying to do is to set the input impedance of gainclone to ~47k because the output impedance of preamp is ~1k. Somewhere on the internet, I've seen that setting the input impedance of an amp to 50 times of output impedance of previous stage is good. BTW, with LM3875T with this configuration, the amp gets very hot. Is it because of low gain or what? I'm sorry for my poor knowledge.
I think you would be alright with an input impedance of 22k with your 1k output from your preamp. With that you can use a 220k-330k feedback resistor and satisfy the >10 gain requirement.
With a 1k output from your preamp, I would use a short interconnect; less than or equal to 1m.
If you want to use the higher impedance, and a pot, you will have to build a non inverting gainclone. Not that this would be the end of the world, but some have stated their inverting clones sounded better. I'm sure the non inverting one will sound better than an inverting one that is oscillating!
The reason your amp was getting hot in the configuration you stated, if I understand it correctly, is that the gain was set up to <10, and the amp was probably oscillating around 10mHz at a couple of watts, driving it absolutely nutty and creating lots of heat.
Follow the rules of the chip, and these troubles will be avoided.
With a 1k output from your preamp, I would use a short interconnect; less than or equal to 1m.
If you want to use the higher impedance, and a pot, you will have to build a non inverting gainclone. Not that this would be the end of the world, but some have stated their inverting clones sounded better. I'm sure the non inverting one will sound better than an inverting one that is oscillating!
The reason your amp was getting hot in the configuration you stated, if I understand it correctly, is that the gain was set up to <10, and the amp was probably oscillating around 10mHz at a couple of watts, driving it absolutely nutty and creating lots of heat.
Follow the rules of the chip, and these troubles will be avoided.
oh yeah, if at all possible, try to have a 'scope handy when working with these chips. they just LOVE to oscillate, so you really gotta keep your eyes open for that. if it's running hot at idle it's most definitely oscillating (unless it's shorted of course). even if it's not particularly hot there may be some low-level oscillations still.
As stated im another thread:
"A lightbulb in series with a 100nF -> 1uF capacitor on the output works. The cap blocks any DC content in the output but AC oscillations pass straight through lighting the bulb for all to see.
Obviously, you need to size the bulb to your power rails.... A 12v car bulb with a series resistor would work."
/micke
"A lightbulb in series with a 100nF -> 1uF capacitor on the output works. The cap blocks any DC content in the output but AC oscillations pass straight through lighting the bulb for all to see.
Obviously, you need to size the bulb to your power rails.... A 12v car bulb with a series resistor would work."
/micke
transformer considerations
I had a few questions about the transformer for my 2 channel gainclone. I was thinking of using a 400VA toroid transformer, and I was wondering about the secondaries' values.
The typical choices are 18VAC and 24VAC.. So that means ~25V and ~34V rails, respectively.. right? Are the different rail voltages choosen to correspond with the speaker impedances? Maybe 18VAC for 4 ohm speakers, 24VAC for 8 ohm?
The speaker I will be using will have two w3-871's in parallel, though one of the drivers will have a 1.5mH inductor and 4 ohm resistor in series with it (to help out with the lower mid-range). And if I understand correctly, that means my speaker will have a 4.8 ohm impedance at lower frequences, and approach 8 ohms at higher frequencies. Does that mean I want 24VAC secondaries?
Also, the w3-871's max out at 30W.. Does that mean I don't want my gainclone at 50W?
Any insight would be appreciated.
--Jordan
I had a few questions about the transformer for my 2 channel gainclone. I was thinking of using a 400VA toroid transformer, and I was wondering about the secondaries' values.
The typical choices are 18VAC and 24VAC.. So that means ~25V and ~34V rails, respectively.. right? Are the different rail voltages choosen to correspond with the speaker impedances? Maybe 18VAC for 4 ohm speakers, 24VAC for 8 ohm?
The speaker I will be using will have two w3-871's in parallel, though one of the drivers will have a 1.5mH inductor and 4 ohm resistor in series with it (to help out with the lower mid-range). And if I understand correctly, that means my speaker will have a 4.8 ohm impedance at lower frequences, and approach 8 ohms at higher frequencies. Does that mean I want 24VAC secondaries?
Also, the w3-871's max out at 30W.. Does that mean I don't want my gainclone at 50W?
Any insight would be appreciated.
--Jordan
Re: transformer considerations
Certainly the lower voltage amp will do better into low (and hopefully more reactive) loads. Since most speakers are more reative than most people realize, this is the way i'm going.
Power handling in terms of home hifi is really more of a marketing thing. More power used intelligently is usually safer than too little power. Just turn the amp down when it starts to clip.
dave
JordanG said:
Certainly the lower voltage amp will do better into low (and hopefully more reactive) loads. Since most speakers are more reative than most people realize, this is the way i'm going.
Power handling in terms of home hifi is really more of a marketing thing. More power used intelligently is usually safer than too little power. Just turn the amp down when it starts to clip.
dave
- just to tell you:
This afternoon I took apart one of my old GCs and made another one following the "Inverted path".
I used sub-miniature carbon resistors for feedback (220k) and input series (10k) resistors (unknown origin, perhaps Phillips). Non-inverted input to ground, 100k lin carbon input pot and a input 47yF Nichicon MUSE bipolar elcap. For PSU I used 2 x 1000yF FRAKO elcaps and a Phillips MKT-HQ 1yF between plus and minus, four conventional 8A bridges and a 2 x 24V / 240VA toroidial transformer. No Zobels etc. All wired with thin silver wire.
This was intended as an experiment and is housed in an open aluminium frame. I used fairly cheap parts, because I think that it has to be good anyway...
Construction was done in "tiny" p2p wiring...
There is no hum and the without input the dammed thing is dead quite.
The sound is fantastic, and I truly admit that I did not expect another big step forward. This baby rocks the house ! I seems a bit bass shy at first glance, but maybe it is more right...
I will run it for another couple of hours (the chips are run in anyway) to test it further.
Ok folks: Thanks for this thread and the other ones. It was overdue ( I wanted to start one a year or so ago, but nobody wanted to jump in... )
And a deep kotau to Kuei Wang Yang.
Klaus
This afternoon I took apart one of my old GCs and made another one following the "Inverted path".
I used sub-miniature carbon resistors for feedback (220k) and input series (10k) resistors (unknown origin, perhaps Phillips). Non-inverted input to ground, 100k lin carbon input pot and a input 47yF Nichicon MUSE bipolar elcap. For PSU I used 2 x 1000yF FRAKO elcaps and a Phillips MKT-HQ 1yF between plus and minus, four conventional 8A bridges and a 2 x 24V / 240VA toroidial transformer. No Zobels etc. All wired with thin silver wire.
This was intended as an experiment and is housed in an open aluminium frame. I used fairly cheap parts, because I think that it has to be good anyway...
Construction was done in "tiny" p2p wiring...
There is no hum and the without input the dammed thing is dead quite.
The sound is fantastic, and I truly admit that I did not expect another big step forward. This baby rocks the house ! I seems a bit bass shy at first glance, but maybe it is more right...
I will run it for another couple of hours (the chips are run in anyway) to test it further.
Ok folks: Thanks for this thread and the other ones. It was overdue ( I wanted to start one a year or so ago, but nobody wanted to jump in... )
And a deep kotau to Kuei Wang Yang.
Klaus
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.