Owen, good stuff, you're all over it. I wonder if they (LCBO) (our gov. run liquor store) take Paypal and offer pickup &/or delivery? 
Still looking for the shopping cart.
I see lots of stock in Ottawa, which one Nepean? and you have a Paypal account? I am a sport, I'd follow through on my offer, its only a silly beer or 2 or ...
TRIPEL KARMELIET
LCBO 287490 | 330 mL bottle
Price: $ 3.00
The Quantity on Hand displayed is as of the Inventory Date indicated. Please contact the store to confirm quantity.
Rick
Still looking for the shopping cart.I see lots of stock in Ottawa, which one Nepean? and you have a Paypal account? I am a sport, I'd follow through on my offer, its only a silly beer or 2 or ...
TRIPEL KARMELIET
LCBO 287490 | 330 mL bottle
Price: $ 3.00
The Quantity on Hand displayed is as of the Inventory Date indicated. Please contact the store to confirm quantity.
Rick
Last edited:
Yeah, my first DIY project in last 20 years.
Finished one channel and powered up. At the moment bias is 300mA, will do some tests with higher bias later, to see if there will be danger of overheating. I'm using heatsinks and power supply from old soviet-time Amfiton 50-YM104 amp, not sure how much power these heatsinks can handle.
At the moment LME and fets are powered by +-33V unregulated. Hooked up my old-timer Korvet 35AC-028 speaker. The sound was far better than i could expect. Have tried the same source and preamp (Amfiton YP-003) with different end-amps and same speakers before, i was pretty sure these speakers are crap. Have to rethink regarding speakers, they sound descently good with theWire amp.
Have to say, design of PCB is excellent, it's doable even if not having any experience with SMD parts. Assembly manual was a great help also.
wb Margus
Finished one channel and powered up. At the moment bias is 300mA, will do some tests with higher bias later, to see if there will be danger of overheating. I'm using heatsinks and power supply from old soviet-time Amfiton 50-YM104 amp, not sure how much power these heatsinks can handle.
At the moment LME and fets are powered by +-33V unregulated. Hooked up my old-timer Korvet 35AC-028 speaker. The sound was far better than i could expect. Have tried the same source and preamp (Amfiton YP-003) with different end-amps and same speakers before, i was pretty sure these speakers are crap. Have to rethink regarding speakers, they sound descently good with theWire amp.
Have to say, design of PCB is excellent, it's doable even if not having any experience with SMD parts. Assembly manual was a great help also.
wb Margus
Hi Guys,
Glad to hear more and more people are getting this done! I'd love to see more pictures if people get the chance.
I'm currently looking at building up my messy implementation in a proper chassis, so I'll be posting some pictures of that endeavor over the next little while.
rsavas:
Sorry for leaving you hanging... for some reason I didn't get a notification for your reply, but I did get one this morning for the note from kiwik.
Your offer for the beer is very kind, but no worries... as a happy side effect of our talk, I ended up checking stock on that beer and found that they did indeed have it in stock at the store near my work, so I quickly ran out at lunch and grabbed a dozen of them
kiwik:
Your implementation sounds really neat... you should post some pictures if you get a chance!
At +/- 33V rails, you should be more than safe to turn the bias up a little unless the heatsinks are really really tiny.
If the heatsink and power supply are up for it, I would push for around 600mA and see how things look. That's only about 40W of dissipation per channel which should be manageable, and basically puts you in class A operation for a good majority of typical listening.
Cheers,
Owen
Glad to hear more and more people are getting this done! I'd love to see more pictures if people get the chance.
I'm currently looking at building up my messy implementation in a proper chassis, so I'll be posting some pictures of that endeavor over the next little while.
rsavas:
Sorry for leaving you hanging... for some reason I didn't get a notification for your reply, but I did get one this morning for the note from kiwik.
Your offer for the beer is very kind, but no worries... as a happy side effect of our talk, I ended up checking stock on that beer and found that they did indeed have it in stock at the store near my work, so I quickly ran out at lunch and grabbed a dozen of them
kiwik:
Your implementation sounds really neat... you should post some pictures if you get a chance!
At +/- 33V rails, you should be more than safe to turn the bias up a little unless the heatsinks are really really tiny.
If the heatsink and power supply are up for it, I would push for around 600mA and see how things look. That's only about 40W of dissipation per channel which should be manageable, and basically puts you in class A operation for a good majority of typical listening.
Cheers,
Owen
kiwik:
Your implementation sounds really neat... you should post some pictures if you get a chance!
At +/- 33V rails, you should be more than safe to turn the bias up a little unless the heatsinks are really really tiny.
If the heatsink and power supply are up for it, I would push for around 600mA and see how things look. That's only about 40W of dissipation per channel which should be manageable, and basically puts you in class A operation for a good majority of typical listening.
Cheers,
Owen[/QUOTE]
Thanks for advice!
Heatsink itself is probably big enough, i'm not sure if thermal conductance between fets and sink is sufficient.
Still have a plan to use separate PSU for LME-s, with some 40-45V voltage. But it's over my skills to build one from scrach, if anyone have a good suggestion where i can find one?
Amp is all over my table now, will take some pics when heatsink is mounted to the frame.
wb Margus
Your implementation sounds really neat... you should post some pictures if you get a chance!
At +/- 33V rails, you should be more than safe to turn the bias up a little unless the heatsinks are really really tiny.
If the heatsink and power supply are up for it, I would push for around 600mA and see how things look. That's only about 40W of dissipation per channel which should be manageable, and basically puts you in class A operation for a good majority of typical listening.
Cheers,
Owen[/QUOTE]
Thanks for advice!
Heatsink itself is probably big enough, i'm not sure if thermal conductance between fets and sink is sufficient.
Still have a plan to use separate PSU for LME-s, with some 40-45V voltage. But it's over my skills to build one from scrach, if anyone have a good suggestion where i can find one?
Amp is all over my table now, will take some pics when heatsink is mounted to the frame.
wb Margus
Potentially stupid question alert ...
Just wanted to check something ...
1. bias trimmer at 5k = 1A bias current (with 10V between pin 11 and 12 on LME chip, 2mA flowing through 5kR ...)
2. bias trimmer at 0R = 0 bias current (with 0V between pin 11 and 12 on LME chip)
linear in between these points?
Just wanted to check something ...
1. bias trimmer at 5k = 1A bias current (with 10V between pin 11 and 12 on LME chip, 2mA flowing through 5kR ...)
2. bias trimmer at 0R = 0 bias current (with 0V between pin 11 and 12 on LME chip)
linear in between these points?
Last edited:
They look really nice, I use a similar one from Coldamp and it's first rate - absolutely no hum at all, which is fab. Greatly improves the amps bass as well of course. At that price you could use one for the LME and one for the output.
Owen I need some catalogue help. My first implimentation of the Lme boards is in an old kenwood Km106 chasis. The rails for the LMe chip are +/-75 and the fet's +/- 68. The lme chip runs hot to touch. ie I can't keep my fingers on the chip/heatsiink for more than a few seconds. I was looking through the Digi-key catalogue and i found some Avid heatsinks similar to the ones you used but about twice as tall. Mounting is different for the chip, bolts not spring clip. Using your vast resourses and knoweledge can you recomend an alternative heat sink for me that will fit the board.
For anyone that is interested I am still testing with a high Z load ie 1000ohms, output voltage is 45 volts RMS just before clipping. My tagret is 35vrms with an 8ohm load. The original kenwood was able to put out 38vrms in an 8ohm load. Pretty good for an amp that was rated at only 130Watts.
For anyone that is interested I am still testing with a high Z load ie 1000ohms, output voltage is 45 volts RMS just before clipping. My tagret is 35vrms with an 8ohm load. The original kenwood was able to put out 38vrms in an 8ohm load. Pretty good for an amp that was rated at only 130Watts.
Could you post the part number for those please? What's the °C/W rating?
I need to buy two such heat sinks. I plan to go nearer to ±100 across the LME.
I think this is the heat sink for the LME chip
digikey WA-T247-101E-ND
Length 1.260" (32.00mm)
Width 0.921" (23.40mm)
Height Off Base (Height of Fin) 0.630" (16.00mm)
Power Dissipation @ Temperature Rise 1W @ 20°C
Thermal Resistance @ Forced Air Flow 8°C/W @ 500 LFM
Thermal Resistance @ Natural 7°C/W
I found this one at digikey
Digi-Key Part Number CR517-ND
Length 2.000" (50.80mm)
Width 0.650" (16.51mm)
Height Off Base (Height of Fin) 0.630" (16.00mm)
Power Dissipation @ Temperature Rise 1.5W @ 20°C
Thermal Resistance @ Forced Air Flow 2°C/W @ 800 LFM
looks like it can disapate more heat, except digikey does not have a price or stock
for the current build I will cut up some old motherboard heatsinks and epoxy them onto the side of the supplied heat sink, won't be pretty and hopefully I won't loose any fingers on the table saw
digikey WA-T247-101E-ND
Length 1.260" (32.00mm)
Width 0.921" (23.40mm)
Height Off Base (Height of Fin) 0.630" (16.00mm)
Power Dissipation @ Temperature Rise 1W @ 20°C
Thermal Resistance @ Forced Air Flow 8°C/W @ 500 LFM
Thermal Resistance @ Natural 7°C/W
I found this one at digikey
Digi-Key Part Number CR517-ND
Length 2.000" (50.80mm)
Width 0.650" (16.51mm)
Height Off Base (Height of Fin) 0.630" (16.00mm)
Power Dissipation @ Temperature Rise 1.5W @ 20°C
Thermal Resistance @ Forced Air Flow 2°C/W @ 800 LFM
looks like it can disapate more heat, except digikey does not have a price or stock
for the current build I will cut up some old motherboard heatsinks and epoxy them onto the side of the supplied heat sink, won't be pretty and hopefully I won't loose any fingers on the table saw
oops I put the following in the wrong thread.
I am still doing testing on the boards and found a couple of things. With some old PC motherboard heatsinks cut up and glued to the LME heatsinks. The LME chips now are just warm instead of too hot to touch. The heatsinks in the old Kenwood poweramp I am using appear to be a bit small so I was trying to find a compromise between bias current and too hot to touch standby opperation. 200 ma appears to be about the limit in my situation. With larger heatsinks I could get away with a higher bias, but that's for the next build. I noticed this morning that the bias was a lot less than 200 ma reading I was getting yesterday. I was trying to see what was going on and found that the bias was very slowly creaping up a ma or so every minute or two. I then tried to monitor the gate to gate voltage to see if it was also going up in relation with the bias current. Once the smoke cleared out of the basement I realized putting a meter across the gates caused the amplifier to oscilate, a lot. Thank goodness for 1/4 watt resistors. I am using 1/4 watt 10 ohm resistors in series with the fets to protect them. I can measure one gate to ground without oscilation. I am waiting for the amplifier to cool so I can start measuring again. The other thing, with +/-68 volt rails the amplifier can easily put out 35 volts rms into an 8ohm load. The feedback resistor is 5.49k 1/8 watt and with about 35 volts across it, it is disapating about 1/4 watt, hmmmm, I am using AC coupled single ended version. I guess for the next build I will have to change the resistors.
I am still doing testing on the boards and found a couple of things. With some old PC motherboard heatsinks cut up and glued to the LME heatsinks. The LME chips now are just warm instead of too hot to touch. The heatsinks in the old Kenwood poweramp I am using appear to be a bit small so I was trying to find a compromise between bias current and too hot to touch standby opperation. 200 ma appears to be about the limit in my situation. With larger heatsinks I could get away with a higher bias, but that's for the next build. I noticed this morning that the bias was a lot less than 200 ma reading I was getting yesterday. I was trying to see what was going on and found that the bias was very slowly creaping up a ma or so every minute or two. I then tried to monitor the gate to gate voltage to see if it was also going up in relation with the bias current. Once the smoke cleared out of the basement I realized putting a meter across the gates caused the amplifier to oscilate, a lot. Thank goodness for 1/4 watt resistors. I am using 1/4 watt 10 ohm resistors in series with the fets to protect them. I can measure one gate to ground without oscilation. I am waiting for the amplifier to cool so I can start measuring again. The other thing, with +/-68 volt rails the amplifier can easily put out 35 volts rms into an 8ohm load. The feedback resistor is 5.49k 1/8 watt and with about 35 volts across it, it is disapating about 1/4 watt, hmmmm, I am using AC coupled single ended version. I guess for the next build I will have to change the resistors.
Good feedback!
Good info "multisync".
Good catch on the feeback resistor power dissipation. I was surprised to see that the National app note used a "MFR-25FBF-6K81' resistor (1/4W/1% Metal Film)
My calcs indicate 70Vpeakx0.7071=~50Vrms squared/6810=0.36W. I was going to use a 1206,1/4W resistor, in my layout, but will change to a 1210,1/2W. We have to realize that we are never going to run the amp at full output 100% of the time so, avg. PD will be significantly less. People talk about thermal modulation distortion, so it is better to over size this particular resistor to minimize this effect.
If you have a scope, suggest to use that to make your measurements, at least if it oscillates, you will see it happening.
Wondering about your statement about 'Thank goodness for 1/4 watt resistors. I am using 1/4 watt 10 ohm resistors in series with the fets to protect them". I thought that you are to be using 220/330ohm resistors in the FET gates?
I have done a layout as well, yet to fab, so I will incorporate all the feedback that everyone is feeding back. Still contemplating changing the LME49830 HS from what was spec'd by Owen. Lower temp Silicon = higher reliability.
Cheers
Rick
Good info "multisync".
Good catch on the feeback resistor power dissipation. I was surprised to see that the National app note used a "MFR-25FBF-6K81' resistor (1/4W/1% Metal Film)
My calcs indicate 70Vpeakx0.7071=~50Vrms squared/6810=0.36W. I was going to use a 1206,1/4W resistor, in my layout, but will change to a 1210,1/2W. We have to realize that we are never going to run the amp at full output 100% of the time so, avg. PD will be significantly less. People talk about thermal modulation distortion, so it is better to over size this particular resistor to minimize this effect.
If you have a scope, suggest to use that to make your measurements, at least if it oscillates, you will see it happening.
Wondering about your statement about 'Thank goodness for 1/4 watt resistors. I am using 1/4 watt 10 ohm resistors in series with the fets to protect them". I thought that you are to be using 220/330ohm resistors in the FET gates?
I have done a layout as well, yet to fab, so I will incorporate all the feedback that everyone is feeding back. Still contemplating changing the LME49830 HS from what was spec'd by Owen. Lower temp Silicon = higher reliability.
Cheers
Rick
the 10ohm resistors are in series with the fets to the fet power supply. I am measuring the voltage across these resistors to set the fet bias. It was not the first time the colour code on the resistors has changed to black black black black. Yes I could set the DVM's to read current but then I would have to change the DVM fuses and probably the fets.
I did have a scope on it but at the output, that's how I knew it was oscilating.
The measured output offset is 0.0000V on one channel and 0.0006V on the other.
90% of the time the finished amplifier will be operating in the milli watt range. 9% in the less than 5 watt range and 1% in the "it's time to dust of the speakers and the neighbours are not home" range.
Rick when you layout your boards leave room for much larger LME heatsinks.
I had the amplifier on for several hours today to see how much the bias current changed over a period of time from cold to hot. Last night before I turned it off I set the bias on each channel to 150ma. This morning the bias was 99ma on one channel and 101 ma on the other. I just checked the bias again after this afternoon's champions league games have finished and the bias has crept up to 144 ma on one chanel and 142 ma on the other.
I was touching the various parts to see how warm they got after being on for about 7 hours. I got a pretty good tingle when I put my finger across the tops of the filter caps. Sure enough there was -75 volts on the top of one the lme caps and -65 on the top of the fet caps. I ohmed out the other set of parts I have and the caps measure several megs from either lead to the metal cap hmmmm.
I will probably take the board apart and replace the gain resistors to be on the safe side.
I did have a scope on it but at the output, that's how I knew it was oscilating.
The measured output offset is 0.0000V on one channel and 0.0006V on the other.
90% of the time the finished amplifier will be operating in the milli watt range. 9% in the less than 5 watt range and 1% in the "it's time to dust of the speakers and the neighbours are not home" range.
Rick when you layout your boards leave room for much larger LME heatsinks.
I had the amplifier on for several hours today to see how much the bias current changed over a period of time from cold to hot. Last night before I turned it off I set the bias on each channel to 150ma. This morning the bias was 99ma on one channel and 101 ma on the other. I just checked the bias again after this afternoon's champions league games have finished and the bias has crept up to 144 ma on one chanel and 142 ma on the other.
I was touching the various parts to see how warm they got after being on for about 7 hours. I got a pretty good tingle when I put my finger across the tops of the filter caps. Sure enough there was -75 volts on the top of one the lme caps and -65 on the top of the fet caps. I ohmed out the other set of parts I have and the caps measure several megs from either lead to the metal cap hmmmm.
I will probably take the board apart and replace the gain resistors to be on the safe side.
Your input solidifies my action to put on a much larger HS. The HS on the National app note is not that much greater Tsa, I will check again, these #'s. Puzzled by the varying bias V. Do you reason that the bias changing has to do with the die temp of the LME? Run a fan on the LME and see how much the bias I changes? Seems to be a rather long Temp slew to stabilize?
"I got a pretty good tingle when I put my finger across the tops of the filter caps."
Silly man, touching live HV ckts with your bare fingers. I work with 347Vac, so stand back when hitting the sw.
I put Ohmite FCSL110R050FERCT on my PCB as an option for source r's
"Sure enough there was -75 volts on the top of one the lme caps and -65 on the top of the fet caps" This sounds correct.
"I ohmed out the other set of parts I have and the caps measure several megs from either lead to the metal cap hmmmm." I don't follow you?
Good to here from people ahead of me thx Rick
"I got a pretty good tingle when I put my finger across the tops of the filter caps."
Silly man, touching live HV ckts with your bare fingers. I work with 347Vac, so stand back when hitting the sw.
I put Ohmite FCSL110R050FERCT on my PCB as an option for source r's
"Sure enough there was -75 volts on the top of one the lme caps and -65 on the top of the fet caps" This sounds correct.
"I ohmed out the other set of parts I have and the caps measure several megs from either lead to the metal cap hmmmm." I don't follow you?
Good to here from people ahead of me thx Rick
multisync, is the voltage across the bias setting trimmer (VR5) changing between amp turn on and and warm up?
I would not have expected the bias to change that drastically, even though you are down in an area where thermal changes for the FET will have an impact on the current I'd expect something like a change from 100mA to 105mA ... I haven't built mine yet so I'm going off datasheets only.
I would not have expected the bias to change that drastically, even though you are down in an area where thermal changes for the FET will have an impact on the current I'd expect something like a change from 100mA to 105mA ... I haven't built mine yet so I'm going off datasheets only.
"I ohmed out the other set of parts I have and the caps measure several megs from either lead to the metal cap hmmmm." I don't follow you?
I was measuring the tops of the caps and had -75volts on the metal top. On the other cap I was expecting 0.0 volts but read around 0.34 volts. WTF I thought I had a bad solder joint or something but all the large bypass caps with metal tops C72, C74 C83 C84 were reading about the same ie not 0.000 volts but a few tenths. So I took out the other caps from the remaining two kits and ohmed out the leads to the metal cap and all the caps read several meg ohms from either lead to the metal cap. So is there a cap expert here who can tell us why the caps ohm out high but have a voltage on them. Never mind I think it is because they are capacitors.
I was monitering the gate voltage but one of my meters died. I went out today and got a fresh batch of 9volt batteries. Before the meter died, the gate voltage did appear to increase over time but not as much. I might have a little bit of coolant spray left so I will try and cool down the lme chip and see what happens.
I am waiting for some fuse holders and banana plugs so I can put this whole thing together and make some noise.
If I decide to change the resistors to a higher value, R43 and R44 are going to be tough for my aging eyesight and shakey fingers. The offset is so low I might leave the input resistors R44, R46 alone and remove the shorting jumper and put a resistor in its place and then change R39 on the bottomn.
I was measuring the tops of the caps and had -75volts on the metal top. On the other cap I was expecting 0.0 volts but read around 0.34 volts. WTF I thought I had a bad solder joint or something but all the large bypass caps with metal tops C72, C74 C83 C84 were reading about the same ie not 0.000 volts but a few tenths. So I took out the other caps from the remaining two kits and ohmed out the leads to the metal cap and all the caps read several meg ohms from either lead to the metal cap. So is there a cap expert here who can tell us why the caps ohm out high but have a voltage on them. Never mind I think it is because they are capacitors.
I was monitering the gate voltage but one of my meters died. I went out today and got a fresh batch of 9volt batteries. Before the meter died, the gate voltage did appear to increase over time but not as much. I might have a little bit of coolant spray left so I will try and cool down the lme chip and see what happens.
I am waiting for some fuse holders and banana plugs so I can put this whole thing together and make some noise.
If I decide to change the resistors to a higher value, R43 and R44 are going to be tough for my aging eyesight and shakey fingers. The offset is so low I might leave the input resistors R44, R46 alone and remove the shorting jumper and put a resistor in its place and then change R39 on the bottomn.
I asked opc about that and he said that increasing the value isnt such a good idea, I asked him about it the other day because I was working out what caps I need on my NTD1 IV stage, basically if your source doesnt have low enough source impedance to deal with the 1.2k, youve built the wrong amp.
he was saying its not a mild effect either
he was saying its not a mild effect either
the balanced version uses values of 1.2 and 30k. So I am pretty sure I can find something between 5.4k and 30k that will work. I have a LMe49810 working for a couple of years with 33k as feedback restors and input resistors. I know I will get a bit more of Johnson noise from the slightly larger values. But I don't thing they make 1/2 watt-1 watt 0805 size resistors.