I've got the Modushop 4U 300mm deep for the EFS3-3 setup based on recommendations on this forum and Gianluca of Modushop so that should be fine with all the UMS tapped holes in the right position.
w
Hi @kokkie
The size of the driver heatsink was something that the Wolverine design team debated and studied in detail over a few weeks.
As with all designs there's not always a perfect solution. However we wanted to get as close to that as we could.
We discussed the following.
- The first was space as we were trying to fit everything onto a board as close to the size of the Honey badger as possible while still allowing for a UMS mounting.
- The second was heat, We had to maximize the size of the heatsink and minimize the bias of the drivers without affecting the performance or causing to large of a temperature rise.
We simulated and tested many combinations of driver transistors, driver emitter resistors and heat sinks dimensions while logging temperature and performance data.
Ultimately we also needed a secure way of mounting it. The mil-max pins do a great job of removing any stress on the transistor legs.
This site was also very useful.
https://www.giangrandi.org/electronics/thcalc/thcalc.shtml
So please rest assured that the size of the heatsink is Optimal. I certainly would not make it any smaller.
Attachments
So, I could theoretically run an EFS3-3 at +/-70VDC with 4U heatsinks using:
And, do the MJL4xxx offer performance benefits over something like the 2SC5200/2SA1943 (besides increased SOA)?
- Drivers: MJE15032G/33G
- Output: MJL4281AG/MJL4302AG
And, do the MJL4xxx offer performance benefits over something like the 2SC5200/2SA1943 (besides increased SOA)?
Beautiful build, @fireanimal !
I'm trying to better understand the tradeoffs between 70V and 65V rails. 70V will result in higher dynamic range and lower distortion, but will put out additional heat.
65V allows for better handling of lower impedance loads.
Did I get that right?
Since the Cobra S2 SMPS can do either, I'm just trying to pick the rails that would be most optimal here.
I'm trying to better understand the tradeoffs between 70V and 65V rails. 70V will result in higher dynamic range and lower distortion, but will put out additional heat.
65V allows for better handling of lower impedance loads.
Did I get that right?
Since the Cobra S2 SMPS can do either, I'm just trying to pick the rails that would be most optimal here.
I have no issues with 70v rails and those or the Sanken MT-200 outputs, as far as I am concerned you can never have too much power / headroom. The heat is not an issue. Also there is no performance penalties running higher rails either, it more boils down to if you want to run the bigger outputs or not really.
There's going to be either a current or voltage limit somewhere in the system. For the same power limitation, the higher voltage makes the 8 ohm output higher while the lower rail should result in more current for lower impedance loads. I'm not familiar with how the Cobra SMPS makes the voltage/current trade-off. Also, if you are driving lower impedance loads, then the lower rail voltage moves you further away from the second breakdown limit. Whether any of this matters to you depends on your speakers and the music that you listen to.
As @fireanimal says, more is better. It is also more expensive, both in parts costs and in power costs. Your rail voltage choices are not very far apart, so there is not a big difference in idle power (<10%)
As @fireanimal says, more is better. It is also more expensive, both in parts costs and in power costs. Your rail voltage choices are not very far apart, so there is not a big difference in idle power (<10%)
I appreciate all the input, it's very helpful!
I'm a sucker for higher-efficiency speakers (most of which are 8 ohm), and it's looking more and more like some new Zu's may be my next major upgrade, so all of this is likely textbook as I'll probably hardly ever go over a few watts lol. But, I wanted to check with others so I appreciate the help.
Another question that I'd like to bring up is around the input diff. semis. I currently have about 25 or so BC560B and BC550B that I could likely pull a close match from, but wondering if the higher hFE of the 559C & 549C is preferential to the better noise specs of the 550/560. Or, is the tradeoff pretty negligible?
I'm a sucker for higher-efficiency speakers (most of which are 8 ohm), and it's looking more and more like some new Zu's may be my next major upgrade, so all of this is likely textbook as I'll probably hardly ever go over a few watts lol. But, I wanted to check with others so I appreciate the help.
Another question that I'd like to bring up is around the input diff. semis. I currently have about 25 or so BC560B and BC550B that I could likely pull a close match from, but wondering if the higher hFE of the 559C & 549C is preferential to the better noise specs of the 550/560. Or, is the tradeoff pretty negligible?
High rail voltage doesn't necessarily equal lower distortion unless your comparing them at a level were the amplifier with the lower rail voltage is beginning to clip.
Higher rails will mean higher output power before it clips, which in a sense will have lower distortion than the one with lower rails because it's not clipping for similar power out.
You certainly don't need 70V rails to get low distortion. 70V rails gives you more clean dynamic output power on transients because it's less likely to clip.
I am not sure what you are referring to. The blue PCB is my GND lift and distribution (parts are a bit oversized, but I did not want to risk anything). Apart from that there is only the Cobra SMPS and the wolverine boards.
I am using the same 2.5mm^2 cable for for AC and DC power, which could be misleading (removed the outer shielding)
Power input is a Schurter plug, double fuses, straight into the SMPS (not using the switch of the plug)
I am using the same 2.5mm^2 cable for for AC and DC power, which could be misleading (removed the outer shielding)
Power input is a Schurter plug, double fuses, straight into the SMPS (not using the switch of the plug)
Match the Vbe first. Then get as close as you can on the Hfe. A simple tester will probably be adequate, although it doesn't necessarily test at the anticipated operating point. The Hfe match on the input pair is important to reduce DC offset, but with the trim pot it is not all that critical. Matched pairs in SMD technology are often matched within a couple of mV Vbe and 10% Hfe. The are some with better specs, but they get expensive fast.
I used a simple battery powered transistor tester to match my input stage transistors. The DC offset after adjustment is bounded by +/- 2mv and is usually less than +/- 1 mV, which should be more than good enough. I matched the Vbe exactly and the Hfe within a couple of per cent at whatever operating point was chosen by the microprocessor in the tester.
The noise specs are not all that much better, depending on where the transistors come from. In the ON-Semi case the difference between the 549C and 550C is just a lower limit on the max noise figure. The typical noise figure is the same. The gain difference between the B and C is unlikely to be audible. You might conceivably measure a small difference in the sub PPM range with the lower beta, but you'll have to work hard to see it. The mirror transistors have a base current helper, so it won't matter all that much, and if you match the input pair the higher offset voltages from the lower beta will cancel.
If you just have to know you have the best possible from this circuit, then the higher Hfe is useful. Understand that this circuit and parts specification is a couple of orders of magnitude better than you can reliably hear in a blind test. You can get 550C but not 560C, at least not as newly made parts. I was able to get the 49/59/C, but I would have happily used B parts if I couldn't get them. I guess I'm just not as much a fanatic as I could be
.