warning:
There may be a risk by adding Teddyregs here. Better start to something more simple. Both channels started playing all right at +/-14.2V but then one of them stopped playing: (-) PS went down to -2.3V.
I've tried many combinations and only could get it work for transitory periods. Last time I played it with LM317/337 it worked OK. The other channel, with TeddyRegs works OK.
Maybe the opamp is damaged or maybe the internal protection circuitry gets activated, shunting PS to ground...I don't know; it is too much for my little knowledge. I will change opamp and start again...with an opamp adaptor which is strongly adviced if one is to tweak the PS.
NOTE: TeddyRegs voltage builds-up slowly and this also could play a role here, for example by activating DC offset detection circuit. There is mod to make the V jump quicker, published at pinkfishmedia.
I'll update ASAP.
Regards,
M
There may be a risk by adding Teddyregs here. Better start to something more simple. Both channels started playing all right at +/-14.2V but then one of them stopped playing: (-) PS went down to -2.3V.
I've tried many combinations and only could get it work for transitory periods. Last time I played it with LM317/337 it worked OK. The other channel, with TeddyRegs works OK.
Maybe the opamp is damaged or maybe the internal protection circuitry gets activated, shunting PS to ground...I don't know; it is too much for my little knowledge. I will change opamp and start again...with an opamp adaptor which is strongly adviced if one is to tweak the PS.
NOTE: TeddyRegs voltage builds-up slowly and this also could play a role here, for example by activating DC offset detection circuit. There is mod to make the V jump quicker, published at pinkfishmedia.
I'll update ASAP.
Regards,
M
Suspect opamp
I had a bad solder joint in the feedback loop once. It opened up one day while music was playing. dumped full B- on the output, until the relay kicked out. Qorked perfectly afterwards.
Ninteen volts is a little high, but I ran them at 17 volts in one for many hours. MP did not recommend runnig the opamp over 12 volts bwcause of clipping response. Raising the rails will allow the LM3886 to overload before the LM318. It is better when the overload is determined by the input stage.
George
I had a bad solder joint in the feedback loop once. It opened up one day while music was playing. dumped full B- on the output, until the relay kicked out. Qorked perfectly afterwards.
Ninteen volts is a little high, but I ran them at 17 volts in one for many hours. MP did not recommend runnig the opamp over 12 volts bwcause of clipping response. Raising the rails will allow the LM3886 to overload before the LM318. It is better when the overload is determined by the input stage.
George
bridging myref revC?
Is it possible to bridge two of the revC monoblocks? Obviously one of them needs to be run in inverted mode which would require some circuitboard changes (which is ok). But I am unclear whether it can be done at all given that this is not the standard way of driving the 3886.
Any advice on what would need to be done would be great. I am starting with Mauro's revC(3) diagram of 18/07/2005 which I believe is the basis for the Twistedpear boards. (please disabuse me of this belief if it is incorrect.)
Thanks much!
Is it possible to bridge two of the revC monoblocks? Obviously one of them needs to be run in inverted mode which would require some circuitboard changes (which is ok). But I am unclear whether it can be done at all given that this is not the standard way of driving the 3886.
Any advice on what would need to be done would be great. I am starting with Mauro's revC(3) diagram of 18/07/2005 which I believe is the basis for the Twistedpear boards. (please disabuse me of this belief if it is incorrect.)
Thanks much!
Re: bridging myref revC?
but, before you make that decision, consider the maximum current demand of the speaker you intend using.
Many builders are adopting peak current:
Ipk=Vpk/Nom Impedance/0.35 Apk
Where will this current come from? Certainly not out of a 3886 when in bridge mode driving a severe reactance 8ohm speaker.
Where will the 3886 get it's current from? Can it be sourced from the smoothing caps?
yes,schro20 said:
but, before you make that decision, consider the maximum current demand of the speaker you intend using.
Many builders are adopting peak current:
Ipk=Vpk/Nom Impedance/0.35 Apk
Where will this current come from? Certainly not out of a 3886 when in bridge mode driving a severe reactance 8ohm speaker.
Where will the 3886 get it's current from? Can it be sourced from the smoothing caps?
Re: Re: bridging myref revC?
I am still quite inexperienced with this, hence I don't even necessarily know just how to ask the right questions...
Thanks for you help!
Let me say some more what motivates me. I built a pair of Linkwitz Pluto speakers (http://linkwitzlab.com/Pluto/intro.htm). They are active and drive the tweeter with a single 3886 and the woofer with a pair of bridged 3886. I want to make a new version of the active board with better amplification (and make a number of other changes) and thought, why not use the revC? Hence my question about the bridging. I am just emulating what is going on in the standard Pluto design (see the diagram at http://linkwitzlab.com/Pluto/electronics.htm).AndrewT said:
I am still quite inexperienced with this, hence I don't even necessarily know just how to ask the right questions...
Thanks for you help!
Ok. Right.BrianDonegan said:
What am I missing though? There are all these bridged LM3886 designs and they seem to work fine. Is the revC design not suitable for this? (since it is not the "standard" GC type setup)
Making sure the powersupply can deliver the juice is certainly a requirement. What about the comment that the LM3886 may not be able to deliver. That part I don't understand. Looking at the datasheet it seems to claim that driving 4 ohm is ok (my load is 8 ohm which would then become 4 ohm to each amp).
Hmmmm. I just found an application note from NS (1192) which talks of bridged loads and designs and such. Ok. Homework.
But the fundamental question remains: is the revC design different in this regard from generic 3886 type setups?
peter
The revC design is no different from it's ordinary brethern in terms of passing transient current through the chipamp.
The 3886 is specified to pass between 7.5Apk and 11Apk when cold through a non-reactive load.
You are asking the chipamp to pass transient current through a reactive load when the chip is warm or even hot. National do not specify the limitations for much of this. They do have an SOAR graph but do not say if the Spike works to that graph or just below those SOAR limits and well below those SOAR limits nor do they tell us how the Spike changes with increase in temperature.
So, back to bridging.
The 8ohm load is seen as a 4ohm load on each half of the bridge.
If you are looking for 120W into 8ohms from a bridged pair of 3886 then each 3886 has to be designed to deliver 60W into 4ohms.
If the load is non reactive then peak current is [Vsupply-Vloss] / 4r0.
For a +-35Vdc Vsupply -Vloss~<=31V/4. Ipk ~<=7.75Apk. This is just above the minimum guaranteed peak delivery of a cold 3886. What is the specified 3886 limit when it's hot? I don't know, National keep it a secret. Are they too embarrassed to tell us. But, probably more to do with the origins of these chips.
Now replace the 4r0 with a 4ohm reactive load. Ipk~=7.75/0.35~=22Apk. NO GO.
Three warm chipamps in parallel may cope with this but more likely 4parallel 3886s will be required making an 8chip BPA120 for +-35Vdc into and 8ohm load.
Reducing the supply to +-20Vdc does not solve the current demand problem, but does help:- keeps the chip cooler and reduces the Ipk.
Do you see why you have to make that current demand decision?
BTW, most chipamp builders on the Forum completely ignore these chipamp limitations and go ahead anyway.
There are alternatives.
Abandon a fully integrated chipamp and go to the discrete output stage or go fully discrete. These both better suit high power output if that is what you need.
The 3886 is specified to pass between 7.5Apk and 11Apk when cold through a non-reactive load.
You are asking the chipamp to pass transient current through a reactive load when the chip is warm or even hot. National do not specify the limitations for much of this. They do have an SOAR graph but do not say if the Spike works to that graph or just below those SOAR limits and well below those SOAR limits nor do they tell us how the Spike changes with increase in temperature.
So, back to bridging.
The 8ohm load is seen as a 4ohm load on each half of the bridge.
If you are looking for 120W into 8ohms from a bridged pair of 3886 then each 3886 has to be designed to deliver 60W into 4ohms.
If the load is non reactive then peak current is [Vsupply-Vloss] / 4r0.
For a +-35Vdc Vsupply -Vloss~<=31V/4. Ipk ~<=7.75Apk. This is just above the minimum guaranteed peak delivery of a cold 3886. What is the specified 3886 limit when it's hot? I don't know, National keep it a secret. Are they too embarrassed to tell us. But, probably more to do with the origins of these chips.
Now replace the 4r0 with a 4ohm reactive load. Ipk~=7.75/0.35~=22Apk. NO GO.
Three warm chipamps in parallel may cope with this but more likely 4parallel 3886s will be required making an 8chip BPA120 for +-35Vdc into and 8ohm load.
Reducing the supply to +-20Vdc does not solve the current demand problem, but does help:- keeps the chip cooler and reduces the Ipk.
Do you see why you have to make that current demand decision?
BTW, most chipamp builders on the Forum completely ignore these chipamp limitations and go ahead anyway.
There are alternatives.
Abandon a fully integrated chipamp and go to the discrete output stage or go fully discrete. These both better suit high power output if that is what you need.
RevC is not working...
I finally assembled my RevCs (TwistedPear PCBs 9/20/2005) last weekend, but they don't work...
My problem:
When switching the transformer on, nothing really happens. Relay doesn't start, LED is dark.
When switching it off, nothing happens at first, and after about 5-6 seconds the relay clicks twice (I think it turns on and off) and the LED lightens up for about half a second at the same time.
I checked all resistors, diodes and caps and there was nothing suspicious, except that I soldered C6 in opposite way at first...
I changed C6 then, but the outcome was the same as mentioned above, nothing changed.
Voltages are +/- 36V on the rails (transformer is a +/- 25V one) an all voltages are as required (about 12.2V across ZD 1&2, 23.5V across R1&4).
The voltage between pin 1 and 7 of the LM318 is 24.2V and the LM3886 Voltages are +/- 36V like the rails...
I really don't know what to do now, any suggestions?
Best regards,
Alex
I finally assembled my RevCs (TwistedPear PCBs 9/20/2005) last weekend, but they don't work...
My problem:
When switching the transformer on, nothing really happens. Relay doesn't start, LED is dark.
When switching it off, nothing happens at first, and after about 5-6 seconds the relay clicks twice (I think it turns on and off) and the LED lightens up for about half a second at the same time.
I checked all resistors, diodes and caps and there was nothing suspicious, except that I soldered C6 in opposite way at first...
I changed C6 then, but the outcome was the same as mentioned above, nothing changed.
Voltages are +/- 36V on the rails (transformer is a +/- 25V one) an all voltages are as required (about 12.2V across ZD 1&2, 23.5V across R1&4).
The voltage between pin 1 and 7 of the LM318 is 24.2V and the LM3886 Voltages are +/- 36V like the rails...
I really don't know what to do now, any suggestions?
Best regards,
Alex
Hawaii,
By putting C6 inverse, You probably generated a big DC signal on the output.
Now not always, but sometimes it might burn R11, the 1 ohm ground resistance.
Then if You don't have ground connection, DC will be present on the output, and the relay will not enable..
Control that R11 is not disconnected.
Ciao, George
By putting C6 inverse, You probably generated a big DC signal on the output.
Now not always, but sometimes it might burn R11, the 1 ohm ground resistance.
Then if You don't have ground connection, DC will be present on the output, and the relay will not enable..
Control that R11 is not disconnected.
Ciao, George
Hawaii cross posted (not generally a good idea).
I think think he has a bad LM318, maybe from his earlier mistake.
Our dialog is here:
http://twistedpearaudio.com/forum/default.aspx?g=posts&m=1398#1398
I think think he has a bad LM318, maybe from his earlier mistake.
Our dialog is here:
http://twistedpearaudio.com/forum/default.aspx?g=posts&m=1398#1398