Some interesting posts above about the gate current questions.
I've found some maths related to the mosfets I'm using, 2SK1058 and 2SJ162.
Please feel fre to correct or add if there are errors with this.
The 6V and 12V mentioned is something that can be limited by Zeners across the mosfets or by the internal Zeners. I don't know quite how that voltage is related to the signal input voltage to the amplifier input or to the bias current setting. Please to elucidate if you know 🙂
The gate capacitance on the Alfets might be different.
The example that Hitachi gives is for 100khz with a gate capacitance of 900 pF and a 6 volt drive which = 3.24 milliwatts per device. However if you increase the drive to say 12 volts you will need nearer 13 milliwatts.
Insufficient gate drive causes distortion to start at a lower and lower frequency.
The formula for gate current:
Power into the MOSFET = Freq being used x Input capacitance of device x ((Voltage of gate to source) squared)
Power in = Freq X capacitance x voltage squared
P in = F x Ciss x Vgs²
Per Hitachi data sheet suggested values:
Each device:
For a:
Freq of 100kHz;
Gate capacitance of 900 pF; (Convert into Farads: 900pf = 900 x 10 to the minus twelve Farads)
Gate voltage of 6V;
So:
P in = 100,000kHz x (900 x 10-¹² F) x 6² = 3.24 mW
Total required is 3.24mW x (the number of devices, eg, 8 mosfets) = 25.92 mW or 12.96 mW per driver. Or the equation could use 8 times the input capacitance?
Each device:
For a:
Freq of 100kHz;
Gate capacitance of 900 pF;
Gate voltage of 12V;
So:
P in = 100,000kHz x (900 x 10-¹² F) x 6² = 12.96 mW
Total required is 12.96 mW x (the number of devices, eg, 8 mosfets) = 103.98 mW or 51.84 mW per driver.
To derive the current:
13mW per gate on a single die mosfet.
12v per gate on a single die mosfet.
P = IV
0.013 W = I x 12 V
0.013 / 12 = I
= 0.00108 Amps
= 1.08mA per gate.
= 3.25mA per driver transistor for three MOSFETs.
= 4.32mA per driver transistor for four MOSFETSs.
So:
= 6.5 mA for six MOSFETs.
= 8.64mA for eight MOSFETs.
"56mA from the LME49830 should be good for 10 pairs of output devices"; more like 25 pairs of single die Hitachi mosfets? The Gate capacitance and current for the Alfets might be different.
I've found some maths related to the mosfets I'm using, 2SK1058 and 2SJ162.
Please feel fre to correct or add if there are errors with this.
The 6V and 12V mentioned is something that can be limited by Zeners across the mosfets or by the internal Zeners. I don't know quite how that voltage is related to the signal input voltage to the amplifier input or to the bias current setting. Please to elucidate if you know 🙂
The gate capacitance on the Alfets might be different.
The example that Hitachi gives is for 100khz with a gate capacitance of 900 pF and a 6 volt drive which = 3.24 milliwatts per device. However if you increase the drive to say 12 volts you will need nearer 13 milliwatts.
Insufficient gate drive causes distortion to start at a lower and lower frequency.
The formula for gate current:
Power into the MOSFET = Freq being used x Input capacitance of device x ((Voltage of gate to source) squared)
Power in = Freq X capacitance x voltage squared
P in = F x Ciss x Vgs²
Per Hitachi data sheet suggested values:
Each device:
For a:
Freq of 100kHz;
Gate capacitance of 900 pF; (Convert into Farads: 900pf = 900 x 10 to the minus twelve Farads)
Gate voltage of 6V;
So:
P in = 100,000kHz x (900 x 10-¹² F) x 6² = 3.24 mW
Total required is 3.24mW x (the number of devices, eg, 8 mosfets) = 25.92 mW or 12.96 mW per driver. Or the equation could use 8 times the input capacitance?
Each device:
For a:
Freq of 100kHz;
Gate capacitance of 900 pF;
Gate voltage of 12V;
So:
P in = 100,000kHz x (900 x 10-¹² F) x 6² = 12.96 mW
Total required is 12.96 mW x (the number of devices, eg, 8 mosfets) = 103.98 mW or 51.84 mW per driver.
To derive the current:
13mW per gate on a single die mosfet.
12v per gate on a single die mosfet.
P = IV
0.013 W = I x 12 V
0.013 / 12 = I
= 0.00108 Amps
= 1.08mA per gate.
= 3.25mA per driver transistor for three MOSFETs.
= 4.32mA per driver transistor for four MOSFETSs.
So:
= 6.5 mA for six MOSFETs.
= 8.64mA for eight MOSFETs.
"56mA from the LME49830 should be good for 10 pairs of output devices"; more like 25 pairs of single die Hitachi mosfets? The Gate capacitance and current for the Alfets might be different.
Last edited:
Sorry at all for OT
Sorry, I do not know you. Perhaps you were part of the list for dps-600/Ncore. I want to clarify. I want to thank Kostas for not having ever doubted me. is important to clarify that the shop AudioPower and customer service are out of the GB and the service they offer is excellent. the group "Kostas buy group" has purchased the DPS-600 for more NCore when it was not on the shop. yes, I admit that there was a delay but I think someone in the group (sub group) did not know that the DPS-600 was not ready available, however, were all reimbursed by PayPal. after they had the money, I wanted to send to Kostas 6 DPS-600. (he was happy about that). I feel sorry for those left out. but now is available on the shop so there is no delay. I want to add, that these are the first two GB of my life. I do not sell products. and, I am not an individual, my name is linked to a company, so do not think anyone loses money, this independant of Paypal.
good luck for you.
Last edited:
I suppose you had the right to reply Roberto as you were questioned/slighted and i'm glad thats out in the open, since I knew the story from kostas, as would anyone who was paying attention in the thread really.
I did not want to make more of it since I knew it had been sorted out. Now thats out of the way, can we please forget about it?
we have spent a lot of space on PSUs, not just DPS, then group buy posts, now NCORE PSU GB grievances too?
can we please move on? please?
the above is more a preventative measure, clocktower, please leave it out or PM, if you really must reply, save it for PM; not here
I did not want to make more of it since I knew it had been sorted out. Now thats out of the way, can we please forget about it?
we have spent a lot of space on PSUs, not just DPS, then group buy posts, now NCORE PSU GB grievances too?
can we please move on? please?
the above is more a preventative measure, clocktower, please leave it out or PM, if you really must reply, save it for PM; not here
Last edited:
Roberto, clocktower has nothing to do with my GB. I know that much.
Yes, there was a big delay in our DPS600/NCore order, but we finally got our units which work fine (and refund for the ones who didn't want to proceed).
Yes, there was a big delay in our DPS600/NCore order, but we finally got our units which work fine (and refund for the ones who didn't want to proceed).
aha, pure trolling then. I searched my inbox for the name to see if it was related to me and got nothing, so could be from my last computer I guess...last year.
seems hes made some posts in this thread about dps400 (the only person), which he was told to get through the shop, but whether it would work with this amp I dont know.
seems hes made some posts in this thread about dps400 (the only person), which he was told to get through the shop, but whether it would work with this amp I dont know.
Last edited:
Hi,
Kostas, thanks for reply.
A delay may occur or problems that happen when you have a public shop. also for people's character (as in life). but I do not know, a problem that has not been solved. I must say it is good, this collaboration helps to keep out any fraud.
-------------------------------
well, I'm waiting for the arrival of 1206 capacitors (I think today), all is ready. I'm working on specifications for this version for Wire Amp, which will contain (up to where I can) some parts of diagrams with value of components,(same on new datasheet for DPS-600/DA). this can help.
I advice that is better replace caps on board (lme) with polyester 0,22uF250V as MKT. is not good in this case, you have 100uF or >, on rail voltage of lme. (lm317 not want big capacitors at output)
Regards
Kostas, thanks for reply.
A delay may occur or problems that happen when you have a public shop. also for people's character (as in life). but I do not know, a problem that has not been solved. I must say it is good, this collaboration helps to keep out any fraud.
-------------------------------
well, I'm waiting for the arrival of 1206 capacitors (I think today), all is ready. I'm working on specifications for this version for Wire Amp, which will contain (up to where I can) some parts of diagrams with value of components,(same on new datasheet for DPS-600/DA). this can help.
I advice that is better replace caps on board (lme) with polyester 0,22uF250V as MKT. is not good in this case, you have 100uF or >, on rail voltage of lme. (lm317 not want big capacitors at output)
Regards
Component ID not know, I am referring to the electrolytic capacitors on the rail of lme.
or change with low capacitance eg. max 22uF100V
yes C72,C83 I presume opc had no problem as is and his regulator is 317/337 based, so perhaps he can comment. one or the other cap should be removed though, the caps on the output of the 317/337, or the caps on the rails of the LME. what caps do you have on the output of the reg on the dps600 AP2?
From the LM317 Datasheet in reference to capacitors on the output:
"Optional—improves transient response. Output capacitors in the range
of 1μF to 1000μF of aluminum or tantalum electrolytic are commonly used
to provide improved output impedance and rejection of transients."
Since we're drawing relatively small and constant current, I decided on 100uF electrolytic bulk local decoupling and a smaller 0.1uF ceramic for HF local decoupling for the LME on the board itself. Note that these capacitors are technically isolated from the regulator to some extent by the wire leads between the regulator and the PCB.
If you're using a regulated supply, you are free to drastically reduce the value of the electrolytic, but it is not necessary. Here's what National says:
"Although the LM117 is stable with no output capacitors, like
any feedback circuit, certain values of external capacitance
can cause excessive ringing. This occurs with values between
500 pF and 5000 pF. A 1 μF solid tantalum (or 25 μF
aluminum electrolytic) on the output swamps this effect and
insures stability. Any increase of the load capacitance larger
than 10 μF will merely improve the loop stability and output
impedance."
I'll leave the final choices to the discretion of the builder, but I did pick the parts and values on the BOM for a reason 🙂
Cheers,
Owen
"Optional—improves transient response. Output capacitors in the range
of 1μF to 1000μF of aluminum or tantalum electrolytic are commonly used
to provide improved output impedance and rejection of transients."
Since we're drawing relatively small and constant current, I decided on 100uF electrolytic bulk local decoupling and a smaller 0.1uF ceramic for HF local decoupling for the LME on the board itself. Note that these capacitors are technically isolated from the regulator to some extent by the wire leads between the regulator and the PCB.
If you're using a regulated supply, you are free to drastically reduce the value of the electrolytic, but it is not necessary. Here's what National says:
"Although the LM117 is stable with no output capacitors, like
any feedback circuit, certain values of external capacitance
can cause excessive ringing. This occurs with values between
500 pF and 5000 pF. A 1 μF solid tantalum (or 25 μF
aluminum electrolytic) on the output swamps this effect and
insures stability. Any increase of the load capacitance larger
than 10 μF will merely improve the loop stability and output
impedance."
I'll leave the final choices to the discretion of the builder, but I did pick the parts and values on the BOM for a reason 🙂
Cheers,
Owen
yep, as with LDO's the transient response is in direct relation to the output capacitance, which at least with LDOs contains everything from that rail to ground. As I understand it, with the LT chips at least, the only problem you can come up against is that it really shouldnt be a super low impedance cap if combined with long inductive traces or wires, as this can cause ringing. I would have thought the 100u chosen is just fine; as..obviously did you =)
I thought that was best coming from you though 😉
i'll probably omit the ones on the regulator output and just leave the ones on the amp
Last edited:
Hi,
I have received my kit of parts.
I find a problem with the LME chip and the heatsink.
As supplied the Clip clamps at the top of the chip. This is fixed by the length of the heatsink and the length of the clip.
If I look carefully I can see that the clips clamps NOT by the bar across the middle of the clip but by the curved ends of the clamp part. These curved ends touch the LME chip at the two top corners. This is not the best place to clamp the LME chip to it's heatsink. I can see daylight passing between the chip face and the heatsink face.
The curved ends of the clip should reach further down the LME chip such that they engage with the two half round recesses where a bolted clamp would be secured. That then allows the middle bar of the clip to contact the body of the LME chip in the middle, well roughly, it's actually about 20% above the middle.
There is a solution.
Get out your hacksaw.
Saw two slots in the top of the heatsink where the two legs of the clamp enter. These two slots would now allow the clip to slide further down the heatsink so the the clamp reaches to a better location on the LME chip.
The two saw cuts can be slightly too long, no harm done, but aim for about 3mm long/deep. 4mm slots will be OK. 2mm slots would be too short.
Now, when the chip is clamped by the bar across the gap between the two recesses you will find that the lead outs protrude from the bottom of the heatsink sufficiently to pass through the PCB and be soldered effectively on the bottom side of the PCB.
You can try using the clamping clip and heatsink as supplied, but there is a risk that the chip will run hot, due to increased Rth c-s and that would be because of that gap I can see.
I'll repeat.
The LME chip is not intended to be clamped by/at the top two corners of the package.
I have received my kit of parts.
I find a problem with the LME chip and the heatsink.
As supplied the Clip clamps at the top of the chip. This is fixed by the length of the heatsink and the length of the clip.
If I look carefully I can see that the clips clamps NOT by the bar across the middle of the clip but by the curved ends of the clamp part. These curved ends touch the LME chip at the two top corners. This is not the best place to clamp the LME chip to it's heatsink. I can see daylight passing between the chip face and the heatsink face.
The curved ends of the clip should reach further down the LME chip such that they engage with the two half round recesses where a bolted clamp would be secured. That then allows the middle bar of the clip to contact the body of the LME chip in the middle, well roughly, it's actually about 20% above the middle.
There is a solution.
Get out your hacksaw.
Saw two slots in the top of the heatsink where the two legs of the clamp enter. These two slots would now allow the clip to slide further down the heatsink so the the clamp reaches to a better location on the LME chip.
The two saw cuts can be slightly too long, no harm done, but aim for about 3mm long/deep. 4mm slots will be OK. 2mm slots would be too short.
Now, when the chip is clamped by the bar across the gap between the two recesses you will find that the lead outs protrude from the bottom of the heatsink sufficiently to pass through the PCB and be soldered effectively on the bottom side of the PCB.
You can try using the clamping clip and heatsink as supplied, but there is a risk that the chip will run hot, due to increased Rth c-s and that would be because of that gap I can see.
I'll repeat.
The LME chip is not intended to be clamped by/at the top two corners of the package.
after wrestling with mine for an eon...I got the clip to rest over the middle of the hole, which seems fine to me, not ideal, but not a problem either IMO. just press the leads against the side of the plated through holes with your iron while flowing solder on and make sure a little flows up the leg (and I mean a little) then hold it there with tweezers or something while removing the soldering iron.
your solution is fine too of course.
your solution is fine too of course.
Dps-600 use unpolarized 1uF for in out and ref of lm317. this is fast caps.
at input, used 22uF100v 105°. from L-C filter (220uH-220nF).
these values are in agreement to the high frequency (150kHz). I agree in principle with opc and know the datasheet. since LM317 has a direct feedback, large output capacity, do not help the time response.
This must be seen in the fact that we have ripple at high frequency (not 60Hz). the lme output voltage is derived from the main stabilization.
at input, used 22uF100v 105°. from L-C filter (220uH-220nF).
these values are in agreement to the high frequency (150kHz). I agree in principle with opc and know the datasheet. since LM317 has a direct feedback, large output capacity, do not help the time response.
This must be seen in the fact that we have ripple at high frequency (not 60Hz). the lme output voltage is derived from the main stabilization.
Last edited:
Hi Owen,
You did "pick" well. I usually test paralleled small-big cap combinations for high Q resonances by applying square wave (with force and speed) at the cap combination. All to often I find ringing (resonance) but not this time..
Regarding the small heatsink that Andrew is mentioning. I recommend some kind of isolation just under the heatsink where the two grounds are exposed. Especially important if one have power and LME ground separated.
Andrew:
This is going to be a long and arduous process if you don't distinguish between a suggested improvement and a critical design flaw.
Your message makes it sound like everyone's amplifier is going to burst into flames if they don't immediately get their hacksaws out and implement your fix. That's really not the case.
Have you measured the temperature delta on the body of the device with both clip locations (upper 3/4 point and dead centre)? Is there even a difference in temp? If you're going to call something a problem, and claim a solution, then you need to prove that.
Were talking about a worst case dissipation of 2.9W if you happen to run 90V rails, or a more common dissipation of about 1.85W if you're running 65V on the LME.
That means the mounting would have the be over 5C/W worse just to get a 10C difference in temp, and I don't think that's very likely to happen by moving the clip a few mm.
I do welcome the suggestions, and it would probably be a great idea if someone really wants to push the LME rails to their limits, but I don't think it's as critical as you're making it sound.
Regards,
Owen
This is going to be a long and arduous process if you don't distinguish between a suggested improvement and a critical design flaw.
Your message makes it sound like everyone's amplifier is going to burst into flames if they don't immediately get their hacksaws out and implement your fix. That's really not the case.
Have you measured the temperature delta on the body of the device with both clip locations (upper 3/4 point and dead centre)? Is there even a difference in temp? If you're going to call something a problem, and claim a solution, then you need to prove that.
Were talking about a worst case dissipation of 2.9W if you happen to run 90V rails, or a more common dissipation of about 1.85W if you're running 65V on the LME.
That means the mounting would have the be over 5C/W worse just to get a 10C difference in temp, and I don't think that's very likely to happen by moving the clip a few mm.
I do welcome the suggestions, and it would probably be a great idea if someone really wants to push the LME rails to their limits, but I don't think it's as critical as you're making it sound.
Regards,
Owen
I said I can see a gap between the LME chip and the heatsink. There is no metal to metal contact !!!!
What I can see is not up for argument.
The solution can be discussed, argued or whatever.
OPC, you are exaggerating.
I said
What I can see is not up for argument.
The solution can be discussed, argued or whatever.
OPC, you are exaggerating.
I said
I did NOT say
Misquoting does no one any credit.
Last edited:
Hi Anders,
Thanks 🙂 Like I said, there's always a reason...
As for the LME heatsink, you are absolutely correct. If you plan to mess around with GND planes and different grounding arrangements then it is imperative that a small piece of electrical tape be used to isolate the heatsink from the two top layer GND planes that it bridges.
I should add this to the building instructions if it's not already there...
If you are running the standard grounding arrangement with the R54 jumper in place, then it's really not an issue.
Regards,
Owen
Thanks 🙂 Like I said, there's always a reason...
As for the LME heatsink, you are absolutely correct. If you plan to mess around with GND planes and different grounding arrangements then it is imperative that a small piece of electrical tape be used to isolate the heatsink from the two top layer GND planes that it bridges.
I should add this to the building instructions if it's not already there...
If you are running the standard grounding arrangement with the R54 jumper in place, then it's really not an issue.
Regards,
Owen
Andrew,
Are you not using the supplied thermal pad? You need electrical isolation between the LME and the heatsink as the tab is connected to V-.
The thermal pad will take care of that gap.
Regards,
Owen
Are you not using the supplied thermal pad? You need electrical isolation between the LME and the heatsink as the tab is connected to V-.
The thermal pad will take care of that gap.
Regards,
Owen