I'm seaching for answers about why my two gainclones amplifiers sound so different.
The first one as been builded by myself. It use :
-Point to point wiring
-lm3886t, good heatsink
-400va (8 X 50va)
-4 diodes bridges mur 860
-regulated using lt1083
-8000 uf per rail X 4 panasonic FC before the regulation
-47uf panasonic FC after regulation
The voltage after regulation is +- 16.5volts with plenty of current. Everything is high current including switches, everything is dual mono with ps is a seperated enclosure.
On the chips, i use
-vishay s102k 0.05% everywhere
-no zobel
-no capacity under the - input for dc offset
My second gainclone :
-chiamp.com kit pcb
-1 transformer 250va 25vac (arround +-34v dc)
-1 rectification board (2 bridges mur 860)
-10 000uf per rail (two rails)
-snubbers on the PS
-lm3886tf
-small panasonic fc on the chips boards
-zobel at the output
-caps in // with the mute resistor
In fact I think it's the complete chiamp.com kit with all feature installed but with one transformer and two rails only.
My system is a sony scd-1 source, a pair of tx-102 mkIII as a preamp and the speakers are bigfunhorn using lowther dx4 (16ohms).
I was really surprised of hearing such difference between two amplifiers based on the same chips each having more power than I need for my application. My firsts gainclones where using lm3875 chips so at the time of building my regulated lm3886 gainclone I blamed the differences on the chips ... maybe I shouldn't have.
The differences between the two are that the chipamp kit is way less contolled in the bass, it's less neutral, it got a 120hz small noise (consider that I get arround 103db eff at the listening position) but the noise is almost irrelevant. Also, even if the chipamp kit got more power with the 34 volts dc, at high volume the sound become blurry and finaly the highs are a little harder and granular.
Also the chipamp kit amplifier heat up, it's warm but not dangerously hot, where the other never heat at all. Is it the results of running it at +-34vdc compared to +-16.5vdc ...
So my questions are :
where would you think thoses differences are coming from ?
Is it the zobel network, the parts, the voltages, the pcb, the snubber, the regulation, the small caps on - input legs of the opamp, the 10 000 uf used without regulation ... ?
Is the heat really function of the voltages when I only use a watts or two in worst situation ?
Keep in mind that I only need 1-2 watts of power for loud listening.
I will start my investigation soon but I would like to have some feedback of people who builded different versions of thoses amplifiers and have developped knowledge about which parts affect the sound is which way.
The first one as been builded by myself. It use :
-Point to point wiring
-lm3886t, good heatsink
-400va (8 X 50va)
-4 diodes bridges mur 860
-regulated using lt1083
-8000 uf per rail X 4 panasonic FC before the regulation
-47uf panasonic FC after regulation
The voltage after regulation is +- 16.5volts with plenty of current. Everything is high current including switches, everything is dual mono with ps is a seperated enclosure.
On the chips, i use
-vishay s102k 0.05% everywhere
-no zobel
-no capacity under the - input for dc offset
My second gainclone :
-chiamp.com kit pcb
-1 transformer 250va 25vac (arround +-34v dc)
-1 rectification board (2 bridges mur 860)
-10 000uf per rail (two rails)
-snubbers on the PS
-lm3886tf
-small panasonic fc on the chips boards
-zobel at the output
-caps in // with the mute resistor
In fact I think it's the complete chiamp.com kit with all feature installed but with one transformer and two rails only.
My system is a sony scd-1 source, a pair of tx-102 mkIII as a preamp and the speakers are bigfunhorn using lowther dx4 (16ohms).
I was really surprised of hearing such difference between two amplifiers based on the same chips each having more power than I need for my application. My firsts gainclones where using lm3875 chips so at the time of building my regulated lm3886 gainclone I blamed the differences on the chips ... maybe I shouldn't have.
The differences between the two are that the chipamp kit is way less contolled in the bass, it's less neutral, it got a 120hz small noise (consider that I get arround 103db eff at the listening position) but the noise is almost irrelevant. Also, even if the chipamp kit got more power with the 34 volts dc, at high volume the sound become blurry and finaly the highs are a little harder and granular.
Also the chipamp kit amplifier heat up, it's warm but not dangerously hot, where the other never heat at all. Is it the results of running it at +-34vdc compared to +-16.5vdc ...
So my questions are :
where would you think thoses differences are coming from ?
Is it the zobel network, the parts, the voltages, the pcb, the snubber, the regulation, the small caps on - input legs of the opamp, the 10 000 uf used without regulation ... ?
Is the heat really function of the voltages when I only use a watts or two in worst situation ?
Keep in mind that I only need 1-2 watts of power for loud listening.
I will start my investigation soon but I would like to have some feedback of people who builded different versions of thoses amplifiers and have developped knowledge about which parts affect the sound is which way.
franksmith said:
Regulated supply vs. Unregulated supply
+/-16.5V vs. 34V
I don't know what topologies (non-inverting vs. inverting) your two amps are but that is a factor as well.
Yes. For the same amount of output power the amp with the higher rails has to drop all that extra voltage across its output transistors.franksmith said:
If you only need a few watts then I'd use well regulated lower voltage supply rails, like you have already done using the LT1083s.franksmith said:
Hi peranders,
I'm pretty aware of subjective impressions and like the blind testing. Sure it would be easy to pick them up, specialy at high spl. Also the small 120hz would help
I'm not a tweaking freak playing endlessly with cables or spikes ... I don't think thoses differences are at the level of cables changes but more at the level of component changes.
I had pretty much the same comments from 2 differents friends.
I'm pretty aware of subjective impressions and like the blind testing. Sure it would be easy to pick them up, specialy at high spl. Also the small 120hz would help
I'm not a tweaking freak playing endlessly with cables or spikes ... I don't think thoses differences are at the level of cables changes but more at the level of component changes.
I had pretty much the same comments from 2 differents friends.
Thanks for the answers,
My regulated one is non inverting, I believe that the chiamp.com kit is non inverting also.
For the voltages differences, I will use a variable transformer at the input of the +-34vdc one to lower it at the same level of the other one for comparison. It sould be a good start.
After I was planning to remove the zobel to see if it can have bad effect with high impedance drivers.
My regulated one is non inverting, I believe that the chiamp.com kit is non inverting also.
For the voltages differences, I will use a variable transformer at the input of the +-34vdc one to lower it at the same level of the other one for comparison. It sould be a good start.
After I was planning to remove the zobel to see if it can have bad effect with high impedance drivers.
It seems to me you have some unresolved construction problems to deal with before you can make a valid comparison of the sound.
The amp that is getting hot may be oscillating. That would certainly affect the sound because it would generate IM products all over the place. You say there is some audible noise coming from that amp - another symptom of a problem that should be addressed. There should be no audible noise no matter what the operating voltage is. Check the rectifiers in the power supply, grounding, etc.
If the temperature difference is merely due to higher operating voltage and different (Inadequate?) heatsinking, the thermal protection circuits in the chip may be turning on and affecting the sound.
I_F
The amp that is getting hot may be oscillating. That would certainly affect the sound because it would generate IM products all over the place. You say there is some audible noise coming from that amp - another symptom of a problem that should be addressed. There should be no audible noise no matter what the operating voltage is. Check the rectifiers in the power supply, grounding, etc.
If the temperature difference is merely due to higher operating voltage and different (Inadequate?) heatsinking, the thermal protection circuits in the chip may be turning on and affecting the sound.
I_F
Hi I_F,
It's one of the first thing I checked on the amp when I doscovered the small heat.
It doesn't seem to get in oscillation, it's something I had to deal before including when building my phonoclone.
For the noise, on regular efficiency speaker I have absolutly no noise (120hz) that I can hear but with the horns I have to deal with high sensitivity and it's something I can get from the other gainclones I've builded when messing up with grounds.
The gainclone which show some 120hz noise got a ground star configuration that seem ok.
I have to admit that I often run my system without ground at all, else my amplifier is the only ground point.
Something I forgot to say is that my regulated gainclone use different enclosure for transfos / bridges and the amplification section. The gainclone showing some 120hz is all builded in the same aluminium enclosure.
It's one of the first thing I checked on the amp when I doscovered the small heat.
It doesn't seem to get in oscillation, it's something I had to deal before including when building my phonoclone.
For the noise, on regular efficiency speaker I have absolutly no noise (120hz) that I can hear but with the horns I have to deal with high sensitivity and it's something I can get from the other gainclones I've builded when messing up with grounds.
The gainclone which show some 120hz noise got a ground star configuration that seem ok.
I have to admit that I often run my system without ground at all, else my amplifier is the only ground point.
Something I forgot to say is that my regulated gainclone use different enclosure for transfos / bridges and the amplification section. The gainclone showing some 120hz is all builded in the same aluminium enclosure.
My 2 cents opinion,
Comparing 1st amp and 2nd amp,
1. regulated supply, the output resistance of the
regulated is very low, my experience power supply
with less o/p impedance than conventional one with no regulation will produce a better bass response.
2. Less supply voltage hence less bias current
hence less heat disspated hence runs a little cooler.
your 2nd amp runs a little hotter.
3. point to point wiring in most cases produces cleaner
sound that is why most tube DIYers do that. There are only a few connections for chipamp, point to point wiring is really not a big deal.
4. 1st amp has less o/p power due to lower supply
voltage. 2nd amp gives you more power. This is not that important to you as I believe you are running efficient speakers.
cheers
Comparing 1st amp and 2nd amp,
1. regulated supply, the output resistance of the
regulated is very low, my experience power supply
with less o/p impedance than conventional one with no regulation will produce a better bass response.
2. Less supply voltage hence less bias current
hence less heat disspated hence runs a little cooler.
your 2nd amp runs a little hotter.
3. point to point wiring in most cases produces cleaner
sound that is why most tube DIYers do that. There are only a few connections for chipamp, point to point wiring is really not a big deal.
4. 1st amp has less o/p power due to lower supply
voltage. 2nd amp gives you more power. This is not that important to you as I believe you are running efficient speakers.
cheers
Hi ttan98,
Thanks for the answer.
I've tried two thing to get rid of the heat.
1- reduce the bandwith with a small 150pf between + and -
This removed the electric sparks catched by the amplifier but did not reduce the heat.
2 - reduced the voltage to a +- 25 volts instead of +-35 volts
This reduced a little bit the heat but did not remove it.
It still produce heat which is not something I'm use to have on gainclone driving this kind of load.
The two thing improved the overall sound, it's still does not have the neutrality of the regulated gainclone and it still sound a little bit lacking control in the bass and mid-bass compared to the regulated one.
I have to investigate more for the heat production.
I'll try to remove the zobel network at the output and also the small capa on the - leg to ground.
I need to check also for the muting current if there is not a mistake in the resistor value.
kuldeepsingh ;
I have no direct comparison of totaly similar gainclone been point to point and using pcb.
So far the two version using pcb that I've tried seem to be bettered by the point to point ones. Since the topologies and components are different, i can't say if the differences come from the pcb.
Thanks for the answer.
I've tried two thing to get rid of the heat.
1- reduce the bandwith with a small 150pf between + and -
This removed the electric sparks catched by the amplifier but did not reduce the heat.
2 - reduced the voltage to a +- 25 volts instead of +-35 volts
This reduced a little bit the heat but did not remove it.
It still produce heat which is not something I'm use to have on gainclone driving this kind of load.
The two thing improved the overall sound, it's still does not have the neutrality of the regulated gainclone and it still sound a little bit lacking control in the bass and mid-bass compared to the regulated one.
I have to investigate more for the heat production.
I'll try to remove the zobel network at the output and also the small capa on the - leg to ground.
I need to check also for the muting current if there is not a mistake in the resistor value.
kuldeepsingh ;
I have no direct comparison of totaly similar gainclone been point to point and using pcb.
So far the two version using pcb that I've tried seem to be bettered by the point to point ones. Since the topologies and components are different, i can't say if the differences come from the pcb.
Frank
I don't understand your response 1, electric spark!! you should not be getting this, I suspect your LM3886's heatsink is not isolated from the heatsink... There maybe a short somewhere.
The heatsink is tied to -12V because the LM3886's heatsink is at -12V.
This only a suspicion.. you may like to check it out.
I don't understand your response 1, electric spark!! you should not be getting this, I suspect your LM3886's heatsink is not isolated from the heatsink... There maybe a short somewhere.
The heatsink is tied to -12V because the LM3886's heatsink is at -12V.
This only a suspicion.. you may like to check it out.
Hi,
With very sensitive speakers, there is some electic spikes (on off from lights or some apliances) that can get their way to the input of the chip and are audible. Thoses are not coming from the +- VS and the pdf of the lm3886 is explaining how to resolve that.
I didn't had any problem with low to medium sensitivity speakers, but all the gainclone I tried on very efficient speakers did that.
Using some polystirene or silver mica, I can't hear differences in sound quality when using them and it reduce the noise floor to nothing.
The chips are isolated, I always check for continuity when I mount them since my heatsinks are grounded.
Something I forgot to check is the balance between + and - rail. I'll give a look to that.
With very sensitive speakers, there is some electic spikes (on off from lights or some apliances) that can get their way to the input of the chip and are audible. Thoses are not coming from the +- VS and the pdf of the lm3886 is explaining how to resolve that.
I didn't had any problem with low to medium sensitivity speakers, but all the gainclone I tried on very efficient speakers did that.
Using some polystirene or silver mica, I can't hear differences in sound quality when using them and it reduce the noise floor to nothing.
The chips are isolated, I always check for continuity when I mount them since my heatsinks are grounded.
Something I forgot to check is the balance between + and - rail. I'll give a look to that.
I'm coming into this with a fading memory of some LM3886 amps I built a few years ago, but FWIW, I remember they had high quescient current. The heatsinks had to be larger than what you'd expect when using the chips at a low power level. I remember something about 2"x2"x3" with fins, even though my design didn't use the full capability of the part. I wasn't happy with the input impedance and offsets on the chip, so I used another high quality op-amp for the front end and, I think, closed the feedback loop around both devices. I haven't read any of the many posts here concerning the chip, so forgive me if this is all old stuff. Still, if any of it is interesting, I can probably dig up the old schematic.
Regards,
CH
Regards,
CH
Resistor matching
I read this with interest, but without actual experiance from building the GC. I have built many circuits with LM3886, but they were all variants of Mauro Penasa's My_Ref.
At first I tried to match up 1% metal films for the two pairs of 22k and 47K resistors it uses. With my 31/2 digit meter the speced 0.1% was not attainable. The 22K were off the most. The first one did generate a little heat.
Then while digging around a local surplus shop ran across a bin with the 0.05% Vishays. All amplifiers have run extremely cool using this overspeced resistor, and the count is up to four.
BTW, all are driving 15 ohm, silver coiled PM7A's. There is no zorbel, but the LM3886 is run at unity gain in these circuits.
Now I have a paralleled LM3886 circuit for driving low impedance, lower efficiency speakers. But it also works just great dealing with the 40 ohm or so of the PM7A's at 15 -20K Hz. It uses SM 0.1% resistors though.
My thoughts on your question; the two amplifiers are so different that I would never expect them to sound alike. I would run the unregulated version at 25 volts by using a 18 volt transformer. And use the Vishays on the PCB. And chuck the zorbel. And put a 2 ohm resistor in the grounds between power ground and signal gropund. And use the offset reducing capcitor.
I used to sweat offset. But a short time using T amps proved that even a 100 mv dc will not harm Lowthers. I used to think that 5 mv was too much.
OT I had an inadvertant pulse test about a year ago. A bad solder joint opened up the feedback loop and I ended up with 34 volts dc on the output. There is a protection circuit built into the My_Ref and Evolution circuits. It opened up after maybe a half second.
My family thought I had committed suicide with a shotgun. My wife ran upstairs to look for my headless corpse, the kids ran outside in case I had missed.
I was still in a state of shock, both ears ringing. My first tought was I had just lost a 13 year old driver and the new rolled edge cones in the closet were going to get installed.
To my surprise the driver is fine. I was turned around when the happened. So do not know how far the driver moved before the power was disconnected. From seeing what a warped record will do, it had to be beyond the limits.
The ear facing that speaker rang for a week. The other quit after a few hours. I calculated this was maybe 130 dB.
Who said that Lowthers will not play LOUD?
George
I read this with interest, but without actual experiance from building the GC. I have built many circuits with LM3886, but they were all variants of Mauro Penasa's My_Ref.
At first I tried to match up 1% metal films for the two pairs of 22k and 47K resistors it uses. With my 31/2 digit meter the speced 0.1% was not attainable. The 22K were off the most. The first one did generate a little heat.
Then while digging around a local surplus shop ran across a bin with the 0.05% Vishays. All amplifiers have run extremely cool using this overspeced resistor, and the count is up to four.
BTW, all are driving 15 ohm, silver coiled PM7A's. There is no zorbel, but the LM3886 is run at unity gain in these circuits.
Now I have a paralleled LM3886 circuit for driving low impedance, lower efficiency speakers. But it also works just great dealing with the 40 ohm or so of the PM7A's at 15 -20K Hz. It uses SM 0.1% resistors though.
My thoughts on your question; the two amplifiers are so different that I would never expect them to sound alike. I would run the unregulated version at 25 volts by using a 18 volt transformer. And use the Vishays on the PCB. And chuck the zorbel. And put a 2 ohm resistor in the grounds between power ground and signal gropund. And use the offset reducing capcitor.
I used to sweat offset. But a short time using T amps proved that even a 100 mv dc will not harm Lowthers. I used to think that 5 mv was too much.
OT I had an inadvertant pulse test about a year ago. A bad solder joint opened up the feedback loop and I ended up with 34 volts dc on the output. There is a protection circuit built into the My_Ref and Evolution circuits. It opened up after maybe a half second.
My family thought I had committed suicide with a shotgun. My wife ran upstairs to look for my headless corpse, the kids ran outside in case I had missed.
I was still in a state of shock, both ears ringing. My first tought was I had just lost a 13 year old driver and the new rolled edge cones in the closet were going to get installed.
To my surprise the driver is fine. I was turned around when the happened. So do not know how far the driver moved before the power was disconnected. From seeing what a warped record will do, it had to be beyond the limits.
The ear facing that speaker rang for a week. The other quit after a few hours. I calculated this was maybe 130 dB.
Who said that Lowthers will not play LOUD?
George
Just to do a little follow-up on this old subject that is still in my mind.
I did some others gainclones since the original post. Unregulated and using pcb.
The sound between the unregulated and pcb using gainclones is pretty similar. They may have different capacitors/resistors/transfos brands but in the end, the differences are not very big. They are pretty good but still the regulated gainclone (using lt1083) is better.
I also did a change in my transformers configuration in the regulated gainclone so I now have +- 24 volts ac or 34v dc and my regulation is now fixed at 28 volts dc per rail. This give me way more power than the original configuration. I expected a change in sound of the amplifier but it's not really noticeable. This is definitively not where the differences are coming from.
Since it would be pretty hard to modify my regulated gainclone (point to point can be a mess) I will instead build another regulated gainclone using pcb with similar parts.
I hope this will help me figure if the differences are more because of the point-to-point wiring or because of the lt-1083 regulation.
BTW : if you build something around the lt1083, be sure to have at minimum 5-7 volts margin for regulation. The manufacturer say 1.5 volts is enough but the performance of the regulator is pretty bad at 1.5v up to 3v. At 5 volts the regulation is very good.
I did some others gainclones since the original post. Unregulated and using pcb.
The sound between the unregulated and pcb using gainclones is pretty similar. They may have different capacitors/resistors/transfos brands but in the end, the differences are not very big. They are pretty good but still the regulated gainclone (using lt1083) is better.
I also did a change in my transformers configuration in the regulated gainclone so I now have +- 24 volts ac or 34v dc and my regulation is now fixed at 28 volts dc per rail. This give me way more power than the original configuration. I expected a change in sound of the amplifier but it's not really noticeable. This is definitively not where the differences are coming from.
Since it would be pretty hard to modify my regulated gainclone (point to point can be a mess) I will instead build another regulated gainclone using pcb with similar parts.
I hope this will help me figure if the differences are more because of the point-to-point wiring or because of the lt-1083 regulation.
BTW : if you build something around the lt1083, be sure to have at minimum 5-7 volts margin for regulation. The manufacturer say 1.5 volts is enough but the performance of the regulator is pretty bad at 1.5v up to 3v. At 5 volts the regulation is very good.
The screeching is from the resistors. SO, Daniel's Mo Powa! mod.
The small metal resistors are probably causing the screeching from the chipamp.com kit.
I recently ordered that LM3886 kit and found it unseemly fidelity.
The following modifications have repaired it.
Replace the resistors with 1/2 watt carbon 5% 56k resistors--to stop the screechy treble.
This is an interesting fit. Both legs of the resistors will have a curve. One end of the resistor will be in the air--about 45 degrees.
The NFB 56k resistor's "nose" is up at centerpoint so that it stays away from the power caps.
Replace the 680R with 1K per National Semiconductor specs.
Don't use the middle hole. Hook up the 47uf capacitor.
The new 1k resistor from NFB to capacitor, its nose is up (away from the chip), touching its own capacitor, about midway up.
Replace the 22k input load resistor with 56k carbon. Its nose is up at the back, "pointing" at the chip (this is opposite of the 1k above).
Replace the 1k input series resistance with an axial 4.7uf capacitor (blackgate is suggested). Use only 1 hole. Put the positive leg in the hole closest to the chip. This capacitor is nose up, away from the chip, and away from the power caps. It sits firmly on the above mentioned 56k input load resistor--away from the power caps.
This lead contacts to the 56k input load resistor.
Curve the other leg back down the towards the circuit board, but Don't hook it up to the board.
Instead,
put a 2.2k resistor in the other hole (this replaces the function of the small 1k metal resistor that you threw out). Use only 1 hole.
Angle this new 2.2k 5% 1/2 watt carbon resistor along the underside of the new 4.7uf input filter cap.
Make a small bend in the leads of both so that they "nest" together naturally. Apply flux to this joint and solder.
From signal to chip goes: Signal, resistor, cap, chip.
Tilt this inwards, away from the power caps. .
Now the chipamp.com audio signal is running entirely on 1/2 watt 5% carbon resistors. There's no more dc output. There's no more screeching. The power output is nearly double, without clipping. And, you now have the option of choosing an input filter capacitor reviewed as "warm" so that you have a fantastic experience with a . . . finally. . . level frequency response.
Mine is happy on 40+40 volts DC rails. Although my test bench power supply is 1 amp rated 30+30, and that made 50 clean watts. Efficiency!!
Sure the heatsink is going to get warm because we just doubled the operating speed of the chip. That's what improved the resolution to double and why it now sounds like LM1875 on steroids.
Use Artic Ceramique or GC Waldham Type 44 for thermal paste.
*Application notes.
1). Its possible to use the smaller 1/4 watt 5% carbon resistors. This is more convenient because you only have to dogleg one lead. However, the 1/2 watt will sound better because the carbon is employed as a screech remover, so more carbon = less screech and more power.
2). The input filter cap gives you many options for tailoring the sound; however, to have a level frequency response, use a cap reviewed as "warm" or "extra bass" for greatest benefit.
3). If gain adjustment is needed, swap the above, new 2.2k in-series resistance for something larger (to turn down the volume).
4). The above configuration supports A20K and higher volume controls. If you wish to do without a volume control, you may add a carbon 22k resistor from signal in, to ground.
5). A minimum 35+35 volts dc is suggested if you want to add the feature of an audiophile size soundfield per each channel (a wide and deep presentation is where a bright amp really shines).
6). Be prepared for a surprise if you punch a bass boost button on the source. Now nearly up to advertised power output and at full bandwidth at double-speed resolution, the speakers may hit X-max.
The small metal resistors are probably causing the screeching from the chipamp.com kit.
I recently ordered that LM3886 kit and found it unseemly fidelity.
The following modifications have repaired it.
Replace the resistors with 1/2 watt carbon 5% 56k resistors--to stop the screechy treble.
This is an interesting fit. Both legs of the resistors will have a curve. One end of the resistor will be in the air--about 45 degrees.
The NFB 56k resistor's "nose" is up at centerpoint so that it stays away from the power caps.
Replace the 680R with 1K per National Semiconductor specs.
Don't use the middle hole. Hook up the 47uf capacitor.
The new 1k resistor from NFB to capacitor, its nose is up (away from the chip), touching its own capacitor, about midway up.
Replace the 22k input load resistor with 56k carbon. Its nose is up at the back, "pointing" at the chip (this is opposite of the 1k above).
Replace the 1k input series resistance with an axial 4.7uf capacitor (blackgate is suggested). Use only 1 hole. Put the positive leg in the hole closest to the chip. This capacitor is nose up, away from the chip, and away from the power caps. It sits firmly on the above mentioned 56k input load resistor--away from the power caps.
This lead contacts to the 56k input load resistor.
Curve the other leg back down the towards the circuit board, but Don't hook it up to the board.
Instead,
put a 2.2k resistor in the other hole (this replaces the function of the small 1k metal resistor that you threw out). Use only 1 hole.
Angle this new 2.2k 5% 1/2 watt carbon resistor along the underside of the new 4.7uf input filter cap.
Make a small bend in the leads of both so that they "nest" together naturally. Apply flux to this joint and solder.
From signal to chip goes: Signal, resistor, cap, chip.
Tilt this inwards, away from the power caps. .
Now the chipamp.com audio signal is running entirely on 1/2 watt 5% carbon resistors. There's no more dc output. There's no more screeching. The power output is nearly double, without clipping. And, you now have the option of choosing an input filter capacitor reviewed as "warm" so that you have a fantastic experience with a . . . finally. . . level frequency response.
Mine is happy on 40+40 volts DC rails. Although my test bench power supply is 1 amp rated 30+30, and that made 50 clean watts. Efficiency!!
Sure the heatsink is going to get warm because we just doubled the operating speed of the chip. That's what improved the resolution to double and why it now sounds like LM1875 on steroids.
Use Artic Ceramique or GC Waldham Type 44 for thermal paste.
*Application notes.
1). Its possible to use the smaller 1/4 watt 5% carbon resistors. This is more convenient because you only have to dogleg one lead. However, the 1/2 watt will sound better because the carbon is employed as a screech remover, so more carbon = less screech and more power.
2). The input filter cap gives you many options for tailoring the sound; however, to have a level frequency response, use a cap reviewed as "warm" or "extra bass" for greatest benefit.
3). If gain adjustment is needed, swap the above, new 2.2k in-series resistance for something larger (to turn down the volume).
4). The above configuration supports A20K and higher volume controls. If you wish to do without a volume control, you may add a carbon 22k resistor from signal in, to ground.
5). A minimum 35+35 volts dc is suggested if you want to add the feature of an audiophile size soundfield per each channel (a wide and deep presentation is where a bright amp really shines).
6). Be prepared for a surprise if you punch a bass boost button on the source. Now nearly up to advertised power output and at full bandwidth at double-speed resolution, the speakers may hit X-max.
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.