input diff biasing zeners ...
well, there is another excellent source of info on this matter:
http://users.ece.gatech.edu/~mleach/lowtim/instage.html
check the second paragraph.
also check the parts list:
http://users.ece.gatech.edu/~mleach/lowtim/part3.html
just trying to help since jens is VERY busy and has certainly done his part
mlloyd1
well, there is another excellent source of info on this matter:
http://users.ece.gatech.edu/~mleach/lowtim/instage.html
check the second paragraph.
also check the parts list:
http://users.ece.gatech.edu/~mleach/lowtim/part3.html
just trying to help since jens is VERY busy and has certainly done his part
mlloyd1
Got my boards, they're massive! Thanks to the guys who made this possible.
I would be glad to have a recipe, too. I would be happy with the missing resistor values for the given output transistor types and, let's say, 60V supply, if that can be worked out by someone with more knowledge in this field than me.
theduke
(d***, they look good!)
I would be glad to have a recipe, too. I would be happy with the missing resistor values for the given output transistor types and, let's say, 60V supply, if that can be worked out by someone with more knowledge in this field than me.
theduke
(d***, they look good!)
Build Questions?
Hi,
Q1. the diode temp comp for the Vbe multiplier can be mounted by accessing the screw hole next to C15 & C27. If the diode PCB is turned over to place the diodes against the heatsink then do we reverse the two wire links (1&2)?
Q2. or do we mount the diode block through access hole next to C2 & C21?
Q3. the output and driver transistors have pads both sides of the board. Do we need to solder both sides? I ask because I would prefer to bolt the Trs to the heatsink first and then solder the legs on the top side since bottom side is no longer accessible.
Q4. as q3 above, the top side of the PCB is not accessible under the electrolytics. Is soldering on bottom side only OK?
Q5. all the little holes around the ground planes, are they plated through to link the upper and lower ground planes?
Q6. the drivers T21 & T22 are called up as needing small heatsinks. Could we mount them under the PCB same as the outputs and bolt them direct to main sink? There appears to be enough board space to redrill the access holes for underboard mounting.
Hi,
Q1. the diode temp comp for the Vbe multiplier can be mounted by accessing the screw hole next to C15 & C27. If the diode PCB is turned over to place the diodes against the heatsink then do we reverse the two wire links (1&2)?
Q2. or do we mount the diode block through access hole next to C2 & C21?
Q3. the output and driver transistors have pads both sides of the board. Do we need to solder both sides? I ask because I would prefer to bolt the Trs to the heatsink first and then solder the legs on the top side since bottom side is no longer accessible.
Q4. as q3 above, the top side of the PCB is not accessible under the electrolytics. Is soldering on bottom side only OK?
Q5. all the little holes around the ground planes, are they plated through to link the upper and lower ground planes?
Q6. the drivers T21 & T22 are called up as needing small heatsinks. Could we mount them under the PCB same as the outputs and bolt them direct to main sink? There appears to be enough board space to redrill the access holes for underboard mounting.
The leads to diode board should stay the same, i.e., 1->1 and 2->2. Current must go through the string in one direction. I was wondering what those holes were for! Duh!
If you use the surface mount diode board, mount it under the hole closest to the 1/2 holes to keep the leads to it as short as possible.
The diode are mounted a bit away from the output transistors. Therefore, the thermal response time will be a little delayed. The surface mount type may have a shorter thermal lag time than the bigger type of diode. The big type of diodes may be better used if they contact the board directly. Don't know if you can do that with the surface mount type, without a thermal heat sink insulator pad. (Not trying to nit-pick here, Jens. Thermal dely is inherent with any such scheme, and you do what you have to do with layout. It's still much easier than Leach's. If the semiconductors manufacturers would just build the diodes into the devices themselves... Oh well.) Another possibility is to mount the diode board to the heat sink between the rows of output transistors under the board. This would put them on the hottest spot on the heat sink, and it would have to be mounted before the board, which would require longer leads. Not as easy as Jens arrangement, but thermally better. If done this way, I would bring back the resistor R25 that was originally in Leach's design, as the leads would be longer.
No. The holes are plated through. The pad on one side of the hole is connected to the other.
yes.
I was wondering that myself. I'm guessing they have to do with current steering in the ground plane.
I would NOT put these on the main heat sink. I think their temperature should remain constant. Self states that beta droop in the output transistors affects the driver transistors through non-linear loading of the drivers. He wasn't positive about why, but mentioned Vbe changes in the drivers when modeling distortion seemed to matter. Vbe changes mostly with temperature, as when more current is drawn through them due to beta droop in the output transistors. While Jens has selected output transistors resitant to beta droop, temperature changes on the heat sink can vary, although slowly. Still, I would think keeping the drivers on their own heat sink would be the best way to keep their temperature as constant as possible. I don't think the thermal tracking arrangement is geared to compensation for temperature changes of the drivers. Kapton washers appear to have the best temperature conductance to the heat sinks.
Hi Pooge,
thanks for your early input - I now have some answers.
Have another look at the diode blocks. They appear as the exact match when looked at from the top. But the diodes need to be in contact with the heatsink. To achieve this both types of diode block need to be flipped over and then the hole numbers do not match. My question is worth repeating:-
do we need to swap over the wires to achieve 1 -> 1 and 2 -> 2, not the 2 pin header called up?
or is the orientation of the diodes ready for a pin header connection?
or do we change the diode orientation to allow the 2 pin header to be used?
My thought for heat conduction to the smt diodes :- apply a small patch of kapton adhesive backed tape to the heatsink and then a splodge (good technical term in my Word dictionary) of heatsink compound to allow best heat transmission into the metal end caps of the smt and some conduction into the body of the diodes and then gently bolt them down. Conformable heat pads may be better but I don't have any.
Your idea of locating the diodes underneath C23 & C29 will certainly reduce time lag and will require wires about 80mm long. A previous contributor did some capacitance and frequency calcs to show that Leach's double resistor decoupling was not necessary. I do not recall anyone disagreeing with him.
The 4 diodes are here to correct for 4 Vbe drops, I think Leach is intending that both the driver and output Vbe are compensated.
On music loads in domestic (not PA) duty the outputs will normally be cold or lookwarm, the drivers will be running hotter than this and so the benefit of the main heatsink for the drivers would be advantageous. A large main heatsink will be of benefit for output cooling in very high ambient temps or for continuous heavy loading into low loads (4ohms during a party night). Then the drivers and outputs will run at similar temps.
thanks for your early input - I now have some answers.
Have another look at the diode blocks. They appear as the exact match when looked at from the top. But the diodes need to be in contact with the heatsink. To achieve this both types of diode block need to be flipped over and then the hole numbers do not match. My question is worth repeating:-
do we need to swap over the wires to achieve 1 -> 1 and 2 -> 2, not the 2 pin header called up?
or is the orientation of the diodes ready for a pin header connection?
or do we change the diode orientation to allow the 2 pin header to be used?
My thought for heat conduction to the smt diodes :- apply a small patch of kapton adhesive backed tape to the heatsink and then a splodge (good technical term in my Word dictionary) of heatsink compound to allow best heat transmission into the metal end caps of the smt and some conduction into the body of the diodes and then gently bolt them down. Conformable heat pads may be better but I don't have any.
Your idea of locating the diodes underneath C23 & C29 will certainly reduce time lag and will require wires about 80mm long. A previous contributor did some capacitance and frequency calcs to show that Leach's double resistor decoupling was not necessary. I do not recall anyone disagreeing with him.
The 4 diodes are here to correct for 4 Vbe drops, I think Leach is intending that both the driver and output Vbe are compensated.
On music loads in domestic (not PA) duty the outputs will normally be cold or lookwarm, the drivers will be running hotter than this and so the benefit of the main heatsink for the drivers would be advantageous. A large main heatsink will be of benefit for output cooling in very high ambient temps or for continuous heavy loading into low loads (4ohms during a party night). Then the drivers and outputs will run at similar temps.
AndrewT said:
Look here for placement of diodes.
An externally hosted image should be here but it was not working when we last tested it.
\Jens
Hi Jens,
In the pic you show, it looks like the diodes are mounted on top of the little PCB. Won't the PCB act as an insulator?
It would seem that the diodes could be mounted on the reverse side of the PCB and them just bolt it down to the heatsink with a little kapton tape.
Blessings, Terry
In the pic you show, it looks like the diodes are mounted on top of the little PCB. Won't the PCB act as an insulator?
It would seem that the diodes could be mounted on the reverse side of the PCB and them just bolt it down to the heatsink with a little kapton tape.
Blessings, Terry
AndrewT said:
1-->1 2-->2 no matter which way you do it.
I didn't see that, and I have no basis for disagreeing with it. Leach simply did it to be safe, and it is not hard to do or extra expensive, so why not? But if you've already built and have no problems, I wouldn't lose sleep over it.
There are 3 Vbe drops from the Vbe multiplier for each polarity, one for each driver and output transistor, or 6 total. I'm not sure about the details of the compensation scheme, and whether or not the drivers are factored in, so I don't want to pretend like I know what I'm talking about here. Leach originally used three diodes. He went to four so they could all be placed in a drilled hole in the heat sink and have the leads on both ends of the string come out on the same side, if I recall correctly. I'd have to look back on this. Since Jens provided a circuit board for easier mounting on the heat sinks, only three diodes could be used and the fourth position bridged, if this in fact makes a better compensation scheme. Something to consider for someone in the know?
Temperature increases raise current gain, which can be a good thing if kept in check so thermal runaway is not a problem. My sinks run pretty warm with Self's optimum bias, which measures around 150ma per channel for two pairs of output devices, and I have large heat sinks (but unfortunately, I had to mount them inside the enclosure). The bottom line is understanding how the compensation is designed. I don't know enough about this.
still4given said:
Doing it this way would also allow one to clip the leads on top of the board to prevent them from poking through any insulator to the heat sinks, too. I've never seen kapton tape, in person, so I don't know what it's like. It seems like a Silpad could offer some cushion, though, to keep the diode from cracking as the board is clamped down, while offering a thermal bridge. I think it would be best to get the diodes in intimate contact.
pooge said:
Kapton tape is like very thin cellophane but stronger. I had thought of using small spacers under the PCB at the bolts to aproximate the thickness of the diodes and then just coating the diodes with thermal grease so that they made good contact with the kapton tape. I didn't want too much pressure against them.
Jen has built the ten channel version. I assume the pic is from that amp. Surely he tested this application and found it satifactory. That is why I asked about it.
Blessings, Terry
Terry,
I placed the diode PCB on the main heatsink of my prototypes with some thermal grease between. I noticed that the thermal time constant is bigger this way, but found no evidence of thermal run-away. Doing it this way requires the builder to take some time to let the amp heat while watching the idle current in the output stage and it will require a couple adjustments of the idle current to get it set where you want it. Once the current is set, I have not had any problems with the amp getting too hot.
I actually improved the thermal diode PCBs for the group order. I placed several vias on the PCB to connect the two sides of the PCB better thermally. I have not yet had a chance to test how this actually affects the setting of the bias current in the output stage, but I expect a little better thermal tracking with this improvement.
If you decide that the thermal tracking of this solution is to slow for you, there is always the option of connecting the diodes like on the great original Leach amp. I found this solution the optimal from a thermal point of view, but always thought it was rather difficult to make it look nice.
\Jens
I placed the diode PCB on the main heatsink of my prototypes with some thermal grease between. I noticed that the thermal time constant is bigger this way, but found no evidence of thermal run-away. Doing it this way requires the builder to take some time to let the amp heat while watching the idle current in the output stage and it will require a couple adjustments of the idle current to get it set where you want it. Once the current is set, I have not had any problems with the amp getting too hot.
I actually improved the thermal diode PCBs for the group order. I placed several vias on the PCB to connect the two sides of the PCB better thermally. I have not yet had a chance to test how this actually affects the setting of the bias current in the output stage, but I expect a little better thermal tracking with this improvement.
If you decide that the thermal tracking of this solution is to slow for you, there is always the option of connecting the diodes like on the great original Leach amp. I found this solution the optimal from a thermal point of view, but always thought it was rather difficult to make it look nice.
\Jens
I wonder if they make suitable flat pack diodes in a 220 package or the like that can be bolted to the heat sink like the output transistors, over some Kapton tape. It looks like there's room on the sink between the output transistors even with the larger size.
Perhaps a header or the like can be used for a disconnect to make it easier to remove the boards for whatever reason, without having to remove the diodes. I'd want to make that a really reliable connection, though.
Perhaps a header or the like can be used for a disconnect to make it easier to remove the boards for whatever reason, without having to remove the diodes. I'd want to make that a really reliable connection, though.
Jens,
The thermal diodes are not there just to prevent thermal runaway. Ideally, they maintain bias at the lowest distortion point. Self investigated various placements for thermal sensors, such as on a TO-3 case, on the heat sink, etc. He found disadvantages for both, and ended up finding putting them on the TO-3 case with some thermal insulation tracked the best.
The multiplier basically has a set characteristic. So the position of the temperature sensor is basically a "least worst" situation to make it work within reasonable bounds of compensation. There are more sophisticated options with more complex circuitry, but we really can't go there with the present topology. While you may not have thermal runaway, it would appear best to mount the sensors as close to the output transistors as possible to keep the bias tracking within a minimum range of distortion, which is more trouble than the convenient way you designed, but easily doable with your design, if desired.
The thermal diodes are not there just to prevent thermal runaway. Ideally, they maintain bias at the lowest distortion point. Self investigated various placements for thermal sensors, such as on a TO-3 case, on the heat sink, etc. He found disadvantages for both, and ended up finding putting them on the TO-3 case with some thermal insulation tracked the best.
The multiplier basically has a set characteristic. So the position of the temperature sensor is basically a "least worst" situation to make it work within reasonable bounds of compensation. There are more sophisticated options with more complex circuitry, but we really can't go there with the present topology. While you may not have thermal runaway, it would appear best to mount the sensors as close to the output transistors as possible to keep the bias tracking within a minimum range of distortion, which is more trouble than the convenient way you designed, but easily doable with your design, if desired.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.