I built a pair of same pcb, yes 2n3019's are hot but it is not a indicator of any fault. Heatsinks are hot too.. you have to attach a small piece of heatsink at least, rather using them naked.
I will post pics of my current setup soon.
@Powerflux
Can you prepare an alternative version respecting my suggestion related with power rails route and pads?
Last edited:
i could, but since it it such a minor modification, some graphical editing with MSpaint like in your example should be sufficient and faster than me and eagle printing PDF's...
That's normal: this is "warm" class AB (sort of)
If there are no resistor inversions or mistakes, the power in each transistor should be ~25V x 13mA = 325mW.
Since the Rth(j-a) is 219°C/W, this means the junction is 71.18°C above ambient.
For 50°C ambient, this translates into 121.18°C junction temp, or, taking into account the Rth(j-c) of 35°C/W, a case temp a little under 110°C.
Probably too hot to touch for many people, but not a concern: this is not a plastic case, and whilst it would be unwise to operate near the 200°C abs max rating, temperatures below the eutectic (~170°C) are perfectly safe in the long term.
If you feel uncomfortable with such temperatures (it is completely output-power independent), there are small tube or chimney-like TO39 heatsinks compatible with Powerflux's design.
You can also use alternative transistors having a Cob <15pF and Ft>75MHz (do not exagerate in that direction though) like the BF819: the TO202 acts as an integral cooling clip.
BF458 and many other are also suitable, but they are more or less obsolete.
Anyway, you can leave your 2N3019's as they are: there is no risk or problem, even in the long term.
Just make sure the resistors are OK, and the zeners have the right voltage and are not shorted.
PS:
the proximity effect of the transistors increases the apparent temperature, but at supply voltages of 25V, this is not a concern.
None of my prototypes has heatsinks fitted, the transistors are almost as close as in Powerflux's PCB, and one of them operates at 40V
Last edited:
Here are two examples of cooling clips for TO5/TO39.
One is a very small, tube-like shape. It could be used in any situation, even where there is very little space.
It is marked "THM 1115" (probably Thermalloy).
The other is a corrugated, chimney-like, and is more obtrusive.
Between these two extreme, one can also find medium-sized grooved cylinders.
I have no example to display unfortunately.
One is a very small, tube-like shape. It could be used in any situation, even where there is very little space.
It is marked "THM 1115" (probably Thermalloy).
The other is a corrugated, chimney-like, and is more obtrusive.
Between these two extreme, one can also find medium-sized grooved cylinders.
I have no example to display unfortunately.
Attachments
Last edited:
I am afraid not: even with 25V rails, the quiescent dissipation is already ~=1W, and the dissipation is heavily output-power-dependent.
That's too much for BCP53.
But even if you exclude "newer" BD140's, there are numbers of TO126 alternatives: some 2SAxxx, and 2SBxxx, I think OdysseyBmx opted for one of those, I'll be able to give you some more details next year....
Mickeymoose has based his version on MJE15031 (TO220). That's heavy artillery, but it works provided the VAS is not too oversized: when you combine slow and/or high capacitance VAS + drivers, stability issues begin to appear.
There will be a fix in Mickey's final version, but the simplest option is to use adequately-sized devices for the job.
I used 2sa1930 as driver on my first build without any obvious problem. But I decided to use grey case Philips BD140's after listening comparisons. 2sa1930 did not sound bad, but BD140 seemed to me good control over output transistors. Same as input transistors.. I favoured BC's over 2sa970 as well. These impressions should be totally related with my setup without making any generalization.
@Elvee
Is 2sc3423 a good candidate in place of 2n3019?
@Elvee
Is 2sc3423 a good candidate in place of 2n3019?
Last edited:
In theory, yes.
But....
with their minuscule capacitances and high Ft, they won't tolerate anything less than a perfect layout, ideal drivers, etc.
In short, if you have the means of taming them, they will give outstanding results but in every other instance, they will oscillate madly and uncontrollably.
Definitely not a beginner's option.
Devices having a Cob around 10pF and an Ft of ~100Mhz will give transparent results without troubles.
Yes, they are the ones.
The BD132 is also suitable, but it is probably also discontinued by NXP.
Those 10w generic parts seem too much different than the 25 year old Mullard/Philips 8w BD140 which had Cob of 5.7pF, Ft of 160Mhz. Question: What conveniently TO220 modern ("widely available") part is most similar to the high performance NOS Philips BD140 used in most classic designs?
For an example of the considerations. . .
Thank you for the examples and helpful posts!
I'd like to use Circlophone in a brand new floor standing radio at my house. It is monophonic, superhetrodyne, with a wideband (full range armed with helper woofer and helper tweeter) 4 ohm speaker. It needs a very powerful amplifier, such as possibly 180w to 4R, with specific performance characteristics such as really expansive sound field that will help the large monophonic radio perform correctly. This floor standing radio is my best guess for a solution to the incessant problem of the television having overtaken and effectively blocked the primary listening area.
I also wish to check out the Circlophone in smaller versions, such as for a driver or preamplifier.
Thank you for the inspirations.
Now starts the search for parts?
The 2SA1209 looks like a perfect clone of the original BD140:
http://www.rapidonline.com/pdf/81-0352e.pdf
It even retains the TO126 case, and it is cheap too.
(I didn't test them, I'll try to find a pair to make a test)
180W will require extremely hefty OP transistors.
The Circlophone is optimally suited to a single pair of output devices, and putting out 180W on a single pair is at the very limit of what can be reliably achieved using relatively normal transistors.
Paralleling is certainly not impossible, but it will require balancing resistors, and the result will be rather messy.
Bridging might be an option.
You made a very good jobHappy New Year!
I tried to make some space for the drivers on powerfuxes PCB Design.
Two files are attached: printout.pdf for lasertoner Transfer and a silkscreen.
Have fun ...
I am sure this will encourage many a would-be builder
I like Sanyo devices. My current setup is using 2sd1047c as output transistors after deciding to use plastic case transistors directly soldered on @powerflux pcbs.. They are a bit faster side (15mhz). Their graph curves are outstanding. I'm using them with 330pF for C12 for a while and it seems this value will be persistent. Bass slam, midrange purity, high softness.. all fits my taste. I think I found my output transistors for my Circlophones.
Last edited:
Another dubious flash of highly misguided insight:
I've imagined an advantage of circlophone outside
sampling that could be taken perhaps even further?
The disadvantage of my middle sampling: That it puts
both non-linear current sampling diodes in the global
path. In a way that is not hidden, nor common mode.
Noble loop has to fix whatever single ended garbage
the peon loop might add or subtract to the output in
the process of taking the non-linear measurement.
Circlophone is at least half immune to this. Its hiding
one of the current sensing diodes behind a collector.
But the other is still in plain sight behind an emitter.
If we can make it so that BOTH diodes are hidden by
collectors: They leave behind no single ended garbage
in output voltage phase with the noble loop. Presenting
only the intended quadrature feedback, in common mode
currents phase.
I've imagined an advantage of circlophone outside
sampling that could be taken perhaps even further?
The disadvantage of my middle sampling: That it puts
both non-linear current sampling diodes in the global
path. In a way that is not hidden, nor common mode.
Noble loop has to fix whatever single ended garbage
the peon loop might add or subtract to the output in
the process of taking the non-linear measurement.
Circlophone is at least half immune to this. Its hiding
one of the current sensing diodes behind a collector.
But the other is still in plain sight behind an emitter.
If we can make it so that BOTH diodes are hidden by
collectors: They leave behind no single ended garbage
in output voltage phase with the noble loop. Presenting
only the intended quadrature feedback, in common mode
currents phase.
Last edited: