I don't have such a gap either, but then I don't have those chips!
Owen, can I remind you again that you were going to let me know the total to pay you so you can send me some.
Regarding the thermal pads, are they a special size? If so, you have four spares you could send with the chips?
I just had a look at the parts list on post 1 of this thread and there's no mention of a thermal pad. The link in post 742 there's no photo of a thermal pad but I do seem to recall somewhere you said there is; maybe in the build instructions?
Thanks.
Ah, here it is:
"With the heatsink pressed tight against the PCB, and the LME sitting with the leads flush with the bottom of the PCB through-holes, use a pencil to mark the top of the IC on the heatsink. Next, cut the grey thermal interface material (TIM) to size, and stick it in the correct location on the heatsink by aligning it with the line you just drew. Once the TIM is adhered, hold the LME in place with one hand (aligned with the pencil line) and push the clip down until it touches the top of the LME. Pry the clip upward with a flat screwdriver, and slide it onto the surface of the LME. If you did everything correctly, the clip should align perfectly with the centre of the hole in the LME IC."
Owen, can I remind you again that you were going to let me know the total to pay you so you can send me some.
Regarding the thermal pads, are they a special size? If so, you have four spares you could send with the chips?
I just had a look at the parts list on post 1 of this thread and there's no mention of a thermal pad. The link in post 742 there's no photo of a thermal pad but I do seem to recall somewhere you said there is; maybe in the build instructions?
Thanks.
Ah, here it is:
"With the heatsink pressed tight against the PCB, and the LME sitting with the leads flush with the bottom of the PCB through-holes, use a pencil to mark the top of the IC on the heatsink. Next, cut the grey thermal interface material (TIM) to size, and stick it in the correct location on the heatsink by aligning it with the line you just drew. Once the TIM is adhered, hold the LME in place with one hand (aligned with the pencil line) and push the clip down until it touches the top of the LME. Pry the clip upward with a flat screwdriver, and slide it onto the surface of the LME. If you did everything correctly, the clip should align perfectly with the centre of the hole in the LME IC."
Last edited:
I stuck the sticky backed insulator to the LME chip. The size fits almost exactly. The hole for the To220 mounting misses the metal interface by sufficient margin.
The insulated LME chip now slides nicely as I try to insert it under the clip.
I suppose for higher pressure on the thermal pad, one could trim the pad shape to more closely follow the "T" shape of the metal back. An inverted triangle would achieve that.
The insulated LME chip now slides nicely as I try to insert it under the clip.
I suppose for higher pressure on the thermal pad, one could trim the pad shape to more closely follow the "T" shape of the metal back. An inverted triangle would achieve that.
IanAS, peace offering, PM me your address i'll send you some Keratherm red. its not adhesive backed, but its a superior material
Thanks Qusp, Now you come to mention it, it's the first time I've looked at Keratherm red. Seems pretty good stuff.
So I had a look around and stumbled upon this 532-4180 at Mouser. It's 15W/mK compared to the 8W/mK of the Keratherm. Aluminium oxide ceramic. I don't if it's flexible and will fill into the tiny surface imperfections or if it's hard like Mica and needs grease. Some Googling says " high density, diamond like hardness", I presume that means grease.
Not sure about the hole, if it comes with one or not, is that a problem for this TO-247 metal rear area? I don't know how I'm going to mount these yet. Looks like they come with a hole. They are 2mm thick.
Also found some Beryllium ceramic pads at 222W/mk. Electrically insulating. Hmmm, makes the others look not all that thermally conductive.
Looks like it's Keratherm then, thanks.
Also I just noticed that the LME49830 seems to have more than halved in price since I looked the other week, unless I previously wrote it down wrong.
So I had a look around and stumbled upon this 532-4180 at Mouser. It's 15W/mK compared to the 8W/mK of the Keratherm. Aluminium oxide ceramic. I don't if it's flexible and will fill into the tiny surface imperfections or if it's hard like Mica and needs grease. Some Googling says " high density, diamond like hardness", I presume that means grease.
Not sure about the hole, if it comes with one or not, is that a problem for this TO-247 metal rear area? I don't know how I'm going to mount these yet. Looks like they come with a hole. They are 2mm thick.
Also found some Beryllium ceramic pads at 222W/mk. Electrically insulating. Hmmm, makes the others look not all that thermally conductive.
Looks like it's Keratherm then, thanks.
Also I just noticed that the LME49830 seems to have more than halved in price since I looked the other week, unless I previously wrote it down wrong.
Last edited:
yeah its pretty good, the stuff I have is the 86/82 in 0.25mm its not their top of the range, but the next down, the top of the range is very good and I have some of that too, but its quite soft/fragile and doesnt really stand up to multiple mountings, for example messing around getting the LME mounted.
yes the lme have basically halved in price, as have most of the range. ever since the Ti takeover, well not straight away, one presumes after they ramped up the production using the Ti fabs
much to opc and my own dismay
yes the lme have basically halved in price, as have most of the range. ever since the Ti takeover, well not straight away, one presumes after they ramped up the production using the Ti fabs
much to opc and my own dismay
I have the first PCB completed.
I built up the AC coupled Single Ended version, with a few changes & additions.
It is on heatsoak at the moment. I may get to listening before I go on extended holiday.
The changes:
R54 = 10r 1% 805
R39 = 30k 805
R46 = 30k 805
R43 = 1k2 805
R44 = 1k2 805
C78 = 2u2F 0.2" pin pitch MKT
Are the resistors 0.1% tolerance?
Added RF filter:
soldered an 805 NP0 100pF in parallel with R46, Just laid it on top thus making a stack of two 805. You could stand these up on edge and solder them as a pair to the pads.
I would have liked to fit a higher value probably around 470pF, but I don't have any 805 between 100pF and 100nF.
The Vgs of the two lateral mosFETs differs enormously, but does not adversely affect the output offset. At 5mA Ib N = 0.2Vgs, P = 0.65Vgs, at 300mA Ib N = 0.66Vgs, P = 1.12Vgs.
Running @ +-45Vdc from a 30+30Vac 160VA transformer. LME supply tapped into the FET supply with 3 wires.
Output offset <1mV, generally flickers between -0.0mVdc and -0.1mVdc (incredibly low offset must be down to the LME and the very close matching of the 30k that are supplied by OPC).
Using a 0.6C/W sink and it is about 20C to 25C above ambient.
5turns of VR5 go from 0mA Ib to 300mA Ib. A further 1turn takes Ib to 590mA.
Assembly.
Solder the Zener after the two adjacent 805.
The pads of C73 & C82 do not project far beyond the package footprint.
I got the file out and sharpened my old and damaged chisel point to a sharp point in an attempt to heat the pad. Only after I failed did I discover that I could have scraped away the paint to expose a "bigger pad". The next 3 PCBs will be scraped before assembly and tinned before corrosion sets in.
BTW, leaving the mute pads open circuit, allows the PSU to unmute the amplifier.
If you, like me, prefers a clockwise rotation of an adjustment to increase the relevant parameter, then insert VR5 back to front, i.e. the screw adjuster is located diagonally opposite the indication on the PCB. This allows for fully anticlockwise to equal zero output bias current. Clockwise rotation brings up the Ib. Be careful, there is no slipping clutch at the zero end, at least not as far as I was prepared to twist to find out.
I have probably forgotten lots, but the above list of comment should help allay any doubts you may have.
I built up the AC coupled Single Ended version, with a few changes & additions.
It is on heatsoak at the moment. I may get to listening before I go on extended holiday.
The changes:
R54 = 10r 1% 805
R39 = 30k 805
R46 = 30k 805
R43 = 1k2 805
R44 = 1k2 805
C78 = 2u2F 0.2" pin pitch MKT
Are the resistors 0.1% tolerance?
Added RF filter:
soldered an 805 NP0 100pF in parallel with R46, Just laid it on top thus making a stack of two 805. You could stand these up on edge and solder them as a pair to the pads.
I would have liked to fit a higher value probably around 470pF, but I don't have any 805 between 100pF and 100nF.
The Vgs of the two lateral mosFETs differs enormously, but does not adversely affect the output offset. At 5mA Ib N = 0.2Vgs, P = 0.65Vgs, at 300mA Ib N = 0.66Vgs, P = 1.12Vgs.
Running @ +-45Vdc from a 30+30Vac 160VA transformer. LME supply tapped into the FET supply with 3 wires.
Output offset <1mV, generally flickers between -0.0mVdc and -0.1mVdc (incredibly low offset must be down to the LME and the very close matching of the 30k that are supplied by OPC).
Using a 0.6C/W sink and it is about 20C to 25C above ambient.
5turns of VR5 go from 0mA Ib to 300mA Ib. A further 1turn takes Ib to 590mA.
Assembly.
Solder the Zener after the two adjacent 805.
The pads of C73 & C82 do not project far beyond the package footprint.
I got the file out and sharpened my old and damaged chisel point to a sharp point in an attempt to heat the pad. Only after I failed did I discover that I could have scraped away the paint to expose a "bigger pad". The next 3 PCBs will be scraped before assembly and tinned before corrosion sets in.
BTW, leaving the mute pads open circuit, allows the PSU to unmute the amplifier.
If you, like me, prefers a clockwise rotation of an adjustment to increase the relevant parameter, then insert VR5 back to front, i.e. the screw adjuster is located diagonally opposite the indication on the PCB. This allows for fully anticlockwise to equal zero output bias current. Clockwise rotation brings up the Ib. Be careful, there is no slipping clutch at the zero end, at least not as far as I was prepared to twist to find out.
I have probably forgotten lots, but the above list of comment should help allay any doubts you may have.
Last edited:
9MHz oscillation if the output is not loaded.
Nothing visible except random noise when 50r loads the output.
Is this a problem I have incorporated?
Or is it a sign that the amp needs a Zobel or Thiele Network to ensure it sees a sensible HF load irrespective of what is hung on the speaker terminals?
Nothing visible except random noise when 50r loads the output.
Is this a problem I have incorporated?
Or is it a sign that the amp needs a Zobel or Thiele Network to ensure it sees a sensible HF load irrespective of what is hung on the speaker terminals?
who had doubts? the only people that have posted in here with built amps have either had no problem, or worked through a problem of their own design/application pretty quickly. the thing has been poked and prodded to within an inch of its life.
i'll let opc comment definitively on whether and where youve maybe caused the problem with the mods, but it would seem likely given the documentation and posting has not mentioned the need for a zobel
yes all resistors are 0.1% minimum
the clockwise mod makes good sense, I didnt bother but I understand the motivation
i'll let opc comment definitively on whether and where youve maybe caused the problem with the mods, but it would seem likely given the documentation and posting has not mentioned the need for a zobel
yes all resistors are 0.1% minimum
the clockwise mod makes good sense, I didnt bother but I understand the motivation
Last edited:
Hi Andrew,
Glad you got it up and running! Sounds like the build went pretty well.
You are absolutely correct about the bias adjust pot. It really should be put in the other way for things to make sense. I highlighted this in the build instructions, so hopefully it doesn't catch anyone off-guard, but rotating it is the quick and easy solution.
As for the oscillation, I have carefully checked both standard configurations with a high BW scope (same one used in the cliiping evaluation) and have not seen any oscillation.
My best guess is that your choice of 10 ohms for R54 is the culprit, but I could be wrong. The only time I've experienced instability with this amp was when I didn't populate that resistor, and tried tying all the grounds back to the PSU separately. With that setup, I saw high frequency oscillations that would turn into severe motorboating with no inputs connected. With a 0R in that position, all was well.
Just out of curiosity, what made you choose that value? Are you grounding everything to a single point back on the PSU?
Can you snap a picture of your setup? When I checked for oscillations, I did have the inputs either connected to the AP or grounded, so I suppose if you have some really long floating wires on there that might be the problem.
Cheers,
Owen
Glad you got it up and running! Sounds like the build went pretty well.
You are absolutely correct about the bias adjust pot. It really should be put in the other way for things to make sense. I highlighted this in the build instructions, so hopefully it doesn't catch anyone off-guard, but rotating it is the quick and easy solution.
As for the oscillation, I have carefully checked both standard configurations with a high BW scope (same one used in the cliiping evaluation) and have not seen any oscillation.
My best guess is that your choice of 10 ohms for R54 is the culprit, but I could be wrong. The only time I've experienced instability with this amp was when I didn't populate that resistor, and tried tying all the grounds back to the PSU separately. With that setup, I saw high frequency oscillations that would turn into severe motorboating with no inputs connected. With a 0R in that position, all was well.
Just out of curiosity, what made you choose that value? Are you grounding everything to a single point back on the PSU?
Can you snap a picture of your setup? When I checked for oscillations, I did have the inputs either connected to the AP or grounded, so I suppose if you have some really long floating wires on there that might be the problem.
Cheers,
Owen
The input is shorted with a tiny wire hoop linking across +IN to -IN.
The output is a pair of twisted stranded flex, about 180mm long.
-IN becomes the Signal Return, when R45=0r0
The Gnd at the signal input triple can be separately connected to Main Audio Ground, if I go for stereo implementation. At the moment it's an uncased monoblock, without any connection to PE.
I can apply a solder short across R54 to see what effect that has. This will be done today.
I can also separately power the LME stage and see if that has an effect. This will have to wait until I get back.
The 9MHz varied between 15mVpp to 1.5Vpp depending on the setting of the scope and where I tapped in the scope probe/earth.
The output is a pair of twisted stranded flex, about 180mm long.
-IN becomes the Signal Return, when R45=0r0
The Gnd at the signal input triple can be separately connected to Main Audio Ground, if I go for stereo implementation. At the moment it's an uncased monoblock, without any connection to PE.
I can apply a solder short across R54 to see what effect that has. This will be done today.
I can also separately power the LME stage and see if that has an effect. This will have to wait until I get back.
The 9MHz varied between 15mVpp to 1.5Vpp depending on the setting of the scope and where I tapped in the scope probe/earth.
I shorted out ground linking resistor R54. That does not change the oscillation.
I can see the 9Mhz switching OFF and ON as I connect/disconnect the 50r load resistor.
Not done any waveform testing. That will have to wait.
I can see the 9Mhz switching OFF and ON as I connect/disconnect the 50r load resistor.
Not done any waveform testing. That will have to wait.
The 330uF 16V caps marked
FP, 19Ab, 331, 16
have a red marking across the top.
Is this red marking an indicator for the +ve lead of this electrolytic capacitor?
FP, 19Ab, 331, 16
have a red marking across the top.
Is this red marking an indicator for the +ve lead of this electrolytic capacitor?
C76 & C78 have ~ 1.5mVdc across them on my build.
Went back to the r54.
The third wire (GND) to the LME PSU bypasses this GND to GND resistor. Shorting the resistor must have zero to near zero effect on the oscillation.
Removed the short and removed the third wire of the PSU link to restore R54 as the only connection and the oscillation is still on my build.
Checked the voltage across R54. 7.6mVdc. This is the non equal current draw of the LME chip's +ve & -ve Pins. About 0.76mA. This voltage drop across R54 equals the new value of output offset. The mosFET end of R54 is at the same voltage as SPKR RETURN. The LME end of R54 is at a similar voltage to SPKR output. The small difference here is due to the "real" output offset as seen by the LME.
This non equal +ve & -ve LME current must return to it's own PSU. This is irrespective of whether the LME is supplied by the same PSU as the mosFETs or has a separate PSU, i.e. the LME GND connection must be made to a PSU.
Qusp,
are you confirming that the red marking is the -ve leadout?
Went back to the r54.
The third wire (GND) to the LME PSU bypasses this GND to GND resistor. Shorting the resistor must have zero to near zero effect on the oscillation.
Removed the short and removed the third wire of the PSU link to restore R54 as the only connection and the oscillation is still on my build.
Checked the voltage across R54. 7.6mVdc. This is the non equal current draw of the LME chip's +ve & -ve Pins. About 0.76mA. This voltage drop across R54 equals the new value of output offset. The mosFET end of R54 is at the same voltage as SPKR RETURN. The LME end of R54 is at a similar voltage to SPKR output. The small difference here is due to the "real" output offset as seen by the LME.
This non equal +ve & -ve LME current must return to it's own PSU. This is irrespective of whether the LME is supplied by the same PSU as the mosFETs or has a separate PSU, i.e. the LME GND connection must be made to a PSU.
Qusp,
are you confirming that the red marking is the -ve leadout?
Last edited:
Just as well my C76 only has -1.5mVdc across it.
Probably not worth turning it around.
Why does Nichicon paint with red, when most others paint similar style caps with a black marker for -ve?
Could this be added to the build guide.
Probably not worth turning it around.
Why does Nichicon paint with red, when most others paint similar style caps with a black marker for -ve?
Could this be added to the build guide.
its worse than that, most caps of this type are marked with a similar bar for the +ve side, not -ve. so theyve marked -ve in a colour that normally indicates +ve in a manner that also normally indicates +ve.
the most reliable way to tell is with the lead lengths ie. long lead is +ve. i've come unstuck relying on the markings enough times to not use them for identifiers without first checking the datasheet for that part
to be fair, they do have several other variants of these polymer caps, some marked in blue, some marked in black; only the FP seem to be in Red.
the most reliable way to tell is with the lead lengths ie. long lead is +ve. i've come unstuck relying on the markings enough times to not use them for identifiers without first checking the datasheet for that part
to be fair, they do have several other variants of these polymer caps, some marked in blue, some marked in black; only the FP seem to be in Red.
Hi Roberto I am completely confused as to were we are today with the DSP600 smps I paid for mine some weeks ago but do not know if it/some/any have actually shipped If it has shipped it may be lost but I do not have tracking info so I am worried, If it has NOT shipped that's fine I am in no rush and want it to be tested and checked first with respect to the modifications - Its just not knowing
John Wallis
John Wallis