I use this company.
Printed Circuit Board - PCBs in 24 Hours.
They are good but it takes about 3-4 weeks from time of order until delivery here in California with the cheap shipping option.
I have been told this company is very good and I think they are faster.
Sitopway Technology Limited
Printed Circuit Board - PCBs in 24 Hours.
They are good but it takes about 3-4 weeks from time of order until delivery here in California with the cheap shipping option.
I have been told this company is very good and I think they are faster.
Sitopway Technology Limited
From my experience with the wolverine when first turned on the bias would start off low, change as the driver and pre drivers warmed up, then continued to rise as the main heat sink warmed up. Once the main heat sink begun to get warm and the main OPS bias transistor also started to heat the steady rise of the bias would begin to slow down, then after a while it would start to decrease and settle in on a much lower value.
For example when turned on the voltage across an RE resistor would be at around 15mV, this would then rise up to around 50mV and then decrease back down to around 22mV once stable. This would take around half an hour.
Like that my OPS was thermally stable but the tracking was extremely poor due to the thermal lag between the main transistors and the bias transistor. To solve this, if you look back, you will see I glued a small SMD transistor directly to one of the transistors in the OPS.
In your case, if you put the bias transistor on flying leads and mount it directly on top of one of the OPS transistors this should improve things significantly. With my SMD transistor there is virtually no drift or lag involved the bias remains pretty much rock solid right at switch on and as the main sinks heat up. Your main heat sinks are much MUCH larger than my little things and it probably takes a very long time for any equilibrium to be reached. I'd try the bias transistor on flying leads and see what happens as it is relatively easy to try.
With regards to the drivers getting hot, mine do warm up but not crazily hot. With my wolverine I am using these for the drivers and driver bias. This is a tiny heat sink 19x33mm and under quiescent conditions I can just about keep my fingers on it.
Under 80 volt rails and a 68R resistor you're getting around 20mA through the drivers and pretty much 80 volts across them. This translates to 1.6 watts each for a total of 3.2 watts.
If you look at my tiny little heat sink, of 13 degrees per watt, that would result in a temperature rise of 41.6 degrees. With an ambient temperature of 25 degrees you'd be at below 70 degrees, which isn't really a concern. Yes, inside a case you would definitely want something bigger than my tiny heat sink, but something 40x66mm would have a thermal resistance 4x lower than mine and you'd easily be able to keep your finger on it.
If I am reading the simulations correctly you should not have trouble with the heat on the drivers.
I asked about flying the bias resistor and didn't get an answer. It is only 1-1/4" away from the nearest output on the heatsink. I figure the heatsink is a better representative of the heat than the plastic top of the outputs but maybe I am mistaken.
Like you, my other amps have always shown a slight drop in bias once the heatsink levels out. This amp doesn't do that. As for the drivers, you can keep your fingers on the heatsink but just barely. Are you running 5 pair outputs on your drivers?
Thanks, Terry
Like you, my other amps have always shown a slight drop in bias once the heatsink levels out. This amp doesn't do that. As for the drivers, you can keep your fingers on the heatsink but just barely. Are you running 5 pair outputs on your drivers?
Thanks, Terry
At quiescent, where the amplifier is amplifying nothing and just sitting there, the current through the drivers is set (afaik) by the resistor between them, the number of output pairs in this case doesn't matter as the driver isn't actually having to drive anything.
What happens under dynamic conditions I am unsure, I was under the impression that OS has set the drivers to work under class A which implies that the current they see remains unchanged regardless of output level, in fact the dissipation goes down when you apply a signal as the voltage dropped across them it as its highest when the amplifier is doing nothing.
Try mounting the plastic case directly on top of one of the output transistors, it is a much better idea than through the heat sink. I mean the physical distance between the two might be quite low but you've got a big hunk of metal in the way. With mine it took 30 mins for the bias to stabilise when the bias transistor was mounted on the heat sink, with it mounted to one of the output its stable as soon as the driver and pre drivers warm up.
From what I remember the piece of metal you've got on your drivers is pretty large. With a 68R resistor this should only be running maybe 15 degrees warmer than ambient which you should just be nicely warm to the touch and certainly not scorching. If you're using a 150R something does seem wrong.
OS was right all those posts ago when he said the drivers run cool, they do.
What happens under dynamic conditions I am unsure, I was under the impression that OS has set the drivers to work under class A which implies that the current they see remains unchanged regardless of output level, in fact the dissipation goes down when you apply a signal as the voltage dropped across them it as its highest when the amplifier is doing nothing.
Try mounting the plastic case directly on top of one of the output transistors, it is a much better idea than through the heat sink. I mean the physical distance between the two might be quite low but you've got a big hunk of metal in the way. With mine it took 30 mins for the bias to stabilise when the bias transistor was mounted on the heat sink, with it mounted to one of the output its stable as soon as the driver and pre drivers warm up.
From what I remember the piece of metal you've got on your drivers is pretty large. With a 68R resistor this should only be running maybe 15 degrees warmer than ambient which you should just be nicely warm to the touch and certainly not scorching. If you're using a 150R something does seem wrong.
OS was right all those posts ago when he said the drivers run cool, they do.
What rails are you running. You can do all the simulations you want. I am running +-77V rails and my drivers are anything but cool. I haven't hear from Ranchu for a while but he was intending on running a 150R in his. Hopefully he will chime in and give a report on how that is working out for him.
I will try flying the bias resistor and see how that works out.
Looking at the circuit, Q102 is called the Driver vbe and Q104 the OPS vbe but they both control the base current at the drivers. Am I missing something? The only difference I see is where they are mounted on the heatsinks. Doesn't turning up the bias affect both the drivers and the outputs?
Thanks, Terry
I will try flying the bias resistor and see how that works out.
Looking at the circuit, Q102 is called the Driver vbe and Q104 the OPS vbe but they both control the base current at the drivers. Am I missing something? The only difference I see is where they are mounted on the heatsinks. Doesn't turning up the bias affect both the drivers and the outputs?
Thanks, Terry
DiAna
In the first place it's matter of performance, that is, a lower measurement floor. An AP isn't suitable for sub ppm distortion measurements (though it's much faster and versatile). On the other hand, I can't deny that the price of an AP was also one of the reasons to start this project. BTW, it's still not finished, as I'm only working on it occasionally.
Cheers, E.
Perhaps this post (first sentence) explains more about his attitude.
Hi Steven,
In the first place it's matter of performance, that is, a lower measurement floor. An AP isn't suitable for sub ppm distortion measurements (though it's much faster and versatile). On the other hand, I can't deny that the price of an AP was also one of the reasons to start this project. BTW, it's still not finished, as I'm only working on it occasionally.
Cheers, E.
I am running 45 volt rails. But the point is the simulation in my case matches what I'm getting in reality. ~20mA through the drivers at 45 volts = 0.9 watts per driver, or 1.8 watts total. Given the 13 degree per watt heat sink I am using that's roughly a 26 degree rise, given ambient is 20-25 at the moment that has it sitting between 45-50 degrees. In line with the just about cool enough to always keep my finger on that they are.
Simulations are incredibly effective at predicting steady state DC conditions and this is precisely what an amplifier is under when not amplifying anything.
Under 77 volt rails you're still getting ~20mA through the drivers, so each one will dissipate 1.55 watt. Or 3.1 watts total.
With my tiny heat sink it'd be safe, but too hot too touch.
Transistor.
This wont alter the current flowing through the drivers mind you, but it should give you a much more solid bias. If it doesn't then something is definitely wrong.
The driver vbe mounts to the small heat sink that you say is getting hot, the one with the MJE15032/33s on.
The OPS vbe mounts to the output stage heat sink, or to an output stage transistor directly.
As far as I understand it, this is a dual tracking approach that is intended to provide a more stable biasing scheme. The transistor mounted to the OPS actually does the hard work, but the one mounted to the drivers provides a degree of compensation to stop things getting way out of hand if the main one fails to do its job correctly. I may be wrong on this mind you, but that's what my intuition here (so much for that perhaps!) is telling me.
Increasing the bias does increase the current through through the drivers, but not by a lot. For example the optimum 107mA biasing for these R22 RE resistor amps results in about 20mA through the drivers. If you half this value and under bias the amplifier significantly then you only bring it down to 18mA through the drivers.
You mentioned that you have to lower the bias right back to keep the driver heat sink cool. If you're seeing 5mV across RE then you've getting 17mA through the drivers, which is hardly any different again.
If the bias keeps creeping up with your amplifiers though, how high would it creep? Because this will increase driver dissipation. As I said before, my bias should have been around 22mV, but it went all the way up to 50+mV before coming down again. If yours isn't tracking correctly it could be going pretty high, which will make things get hotter. Just not by that much, unless it's really running away.
I use an EMU 0204 (USB) with excellent results (4-6 ppm THD). It has the benefit of a balaced input which is handy in getting rid of pesky earth loops.
Edmond,
Thank you for the reply. Yes it does appear that Waly has chosen to be obnoxious on purpose, why he is so convinced he is correct appears to only be on technical information with no real world experience to back that up. I have learned long ago that what is in a book can be just a wrong as some subjective opinions, this however was in another field entirely. I did much research and development in applications of new polymer chemistry and what they thought would happen from small scale trials in a lab often didn't work out in reality. Many times we had to do things we were told couldn't work until we handed then the results that proved them wrong.
I guess your software development is a labor or love, it does take time when you are working on many projects and doing it under the radar so to speak. Perhaps one day you will be able to share it with the rest of us! The AP equipment is not cheap, neither is the Kippler laser analysis equipment that would be nice doing my speaker development. I guess I am stuck with using my Clio system for now and working with a friend and his B&K lab equipment.
Thank you for the reply. Yes it does appear that Waly has chosen to be obnoxious on purpose, why he is so convinced he is correct appears to only be on technical information with no real world experience to back that up. I have learned long ago that what is in a book can be just a wrong as some subjective opinions, this however was in another field entirely. I did much research and development in applications of new polymer chemistry and what they thought would happen from small scale trials in a lab often didn't work out in reality. Many times we had to do things we were told couldn't work until we handed then the results that proved them wrong.
I guess your software development is a labor or love, it does take time when you are working on many projects and doing it under the radar so to speak. Perhaps one day you will be able to share it with the rest of us! The AP equipment is not cheap, neither is the Kippler laser analysis equipment that would be nice doing my speaker development. I guess I am stuck with using my Clio system for now and working with a friend and his B&K lab equipment.
I am running 45 volt rails. But the point is the simulation in my case matches what I'm getting in reality. ~20mA through the drivers at 45 volts = 0.9 watts per driver, or 1.8 watts total. Given the 13 degree per watt heat sink I am using that's roughly a 26 degree rise, given ambient is 20-25 at the moment that has it sitting between 45-50 degrees. In line with the just about cool enough to always keep my finger on that they are.
OK, here's the rub. Earlier you said cool. Now you say just cool enough to keep you fingers on them. When I hear someone say cool, that is not the picture I form.
Simulations are incredibly effective at predicting steady state DC conditions and this is precisely what an amplifier is under when not amplifying anything.
Under 77 volt rails you're still getting ~20mA through the drivers, so each one will dissipate 1.55 watt. Or 3.1 watts total.
With my tiny heat sink it'd be safe, but too hot too touch.
With mine you can touch them but they are pretty hot. If being able to keep your fingers on them is the goal I guess I'm there. It's the term "cool" that had me concerned.
Transistor.
Typo
This wont alter the current flowing through the drivers mind you, but it should give you a much more solid bias. If it doesn't then something is definitely wrong.
The driver vbe mounts to the small heat sink that you say is getting hot, the one with the MJE15032/33s on.
I used MJW3281/MJW1302 for these.
The OPS vbe mounts to the output stage heat sink, or to an output stage transistor directly.
As far as I understand it, this is a dual tracking approach that is intended to provide a more stable biasing scheme. The transistor mounted to the OPS actually does the hard work, but the one mounted to the drivers provides a degree of compensation to stop things getting way out of hand if the main one fails to do its job correctly. I may be wrong on this mind you, but that's what my intuition here (so much for that perhaps!) is telling me.
Increasing the bias does increase the current through through the drivers, but not by a lot. For example the optimum 107mA biasing for these R22 RE resistor amps results in about 20mA through the drivers. If you half this value and under bias the amplifier significantly then you only bring it down to 18mA through the drivers.
You mentioned that you have to lower the bias right back to keep the driver heat sink cool. If you're seeing 5mV across RE then you've getting 17mA through the drivers, which is hardly any different again.
No, not to keep the driver bias low, the output bias.
If the bias keeps creeping up with your amplifiers though, how high would it creep? Because this will increase driver dissipation. As I said before, my bias should have been around 22mV, but it went all the way up to 50+mV before coming down again. If yours isn't tracking correctly it could be going pretty high, which will make things get hotter. Just not by that much, unless it's really running away.
I won't say it completely runs away. If I set the bias to 25mV across a pair of emitters it will rise to 60mV (30mV each), which amounts to very hot heatsinks. Way hotter than I want it to run.
Thank, Terry
Have little doubt. Platform controller hub link to CPU seems (4x link) PCIe (vX.X) both Intel or AMD, therefor seems some interfacing/bridging has to take place. Not shure but seems logic that inside platform controller hub is build in PCI-PCIe bridge though it's maybe a better brand or licensed one verse Z-chipset's where placed external. Shorter route than that could be made direct to CPU by PLX PCI-PCIe x1 link.
Wouldn't all this PC talk fit better it it's own thread?
I will try that. I just never had one that didn't seem to settle.
I think everyone's idea of cool is different. Hot to me = too hot to touch. Cool in electrical terms to me = you can touch it indefinitely. Running cold = barely any different from ambient.
~3 watts of dissipation for the combined driver pair isn't a lot but it does require some heat sinking not to get uncomfortable to the touch.
Ah right, that makes more sense
As AJT says though, when setting the bias in an EF amplifier what you need to do is turn it on, let it warm up, then you set your bias to the point you require. In this case 107mA through the RE resistors, so you want to measure a drop of 23.54mV.
Once you've set it to ~23mV you leave the amplifier alone for a while and let it cooldown/warmup again. Then you check the bias. If it has changed, you set it back to 23mV again. Then you leave the amplifier alone. You basically keep on repeating this until you are happy with the result.
The real test is to then turn the amplifier off. Let it cool down. Turn it back on again and see if, after it warms up, it ends up at 23mV again.
If you mount the vbe transistor to a transistor in the output stage on flying leads you generally don't have to go through half as much of this repetition and the amplifier performs better too as a result.
Personally I think the PC talk fits very well in this thread. We are building amplifiers and testing them to make sure they are performing correctly. We are using PCs to do this and some people might be in need of some help on how to actually do this.
If you were having trouble with zobels smoking and the problem was your ability to measure the AC voltage correctly at high frequencies then we'd discuss that here. We did discuss that here your multimeter wasn't quite up to the task but your scope was.
Thimios is trying to measure another aspect of his amplifier and is having some difficulty with his hardware set up. We're trying to help him over come that trouble, no doubt he isn't the only one.
Building something is just as much about confirming that it is performing correctly as it is as much about the building and listening.