Hi Struth
I had a feeling that this layout would meet with a mixed reaction and I accept that it is a compromise (isn't everything?). I have tried other configurations and keep coming back to this so let me explain my rationale...
Aesthetically I prefer a compact chassis with a fully enclosed heatsink, which I feel is reasonable for the modest (~50W) projects I built. I use a single large heatsink with ample ventilation through the top and bottom plates to maximise airflow across the heatsink and minimise internal heat build up. I also like to position the heatsink so that it forms a shield between the transformer/rectifier diodes and the PC boards/signal wiring to minimise EM and RF interference. I have found in practice that maximising the air gap between the transformer and the sensitive electronics contributes to a very quiet amplifier, even though the toroidals available locally seem to vary in quality.
The most space-efficient configuration is to run the PC boards parallel to the heatsink. I use a 2RU chassis with a 250x75mm heatsink squeezed in so this sets the PC board dimensions to 100x74mm, allowing for clearances and mounting hardware.
With the TO-247 devices underslung and the TO-126 bias sense device sandwiched between an output transistor and the PC Board:
- there is good thermal tracking because it is sensing the top of the package rather than the heatsink directly (less thermal lag)
- traces to the sense transistor can be routed on board, without resort to flying leads, which complicates assembly and with their added lead inductances to worry about.
- a second TO-126 transistor is used as a spacer for the other TO-247 output.
- PC board traces can be kept very tight, particularly when SMT devices are used, and I believe that this improves performance. I built a simple VFA to this layout with 30 MHz outputs and > 3.8 MHz intercept. Simulated THD & slew rate were very good and the sound quality wonderful, showing what can be achieved with a simple circuit and optimised layout.
With the air gap around the TO-247s, and ample ventilation above/below the assembly (as there should be anyway) I believe that convection currents do a good job of drawing heat away from the backside of the heatsink. PC board heating, through the TO-126 spacers that are in contact with the outputs, appears minimal in the examples I have seen. I have conducted some 1/3 power testing into a dummy load, in the ventilated enclosure, and then measured temperatures of the heatsink, PC boards and surrounding components. Everything was pretty hot (as you would expect) but the PC boards not much hotter than the components and significantly cooler than the heatsink.
I don't claim that this is the best solution for everyone, and its probably not well-suited to high power circuits, but it has worked well for me with my own modest requirements.
Last edited:
Great! Now you'll know the current in Q1 (= V_across_R6 / 100.0) and the current in Q2 (= V_across_R7 / 100.0) to within 0.1%. This will tell you how imbalanced the input pair is, and lead you towards the optimum ratio of (R10 / R21) and (R11 / R22) which yields perfect balance. You can experiment with it in SPICE beforehand to get a feeling for the sensitivity (%change_of_imbalance / %change_in_R10).
Hi Struth
If you take a closer look you will see that the TO-126 drivers are mounted vertically on the board, and although it is not marked on the silkscreen, there is sufficient PC board clearance to attach a small TO-220 heatsink to each.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.