The same model transistor for both Q13 and Q12, right?
Also checking current capacity of D4, D5, (at least a 5a part) as they do look small but you know I couldn't actually read the labels on the photo.
In fact, the BF819 have been tested, and they work:
http://www.diyaudio.com/forums/solid-state/189599-my-little-cheap-circlophone-51.html#post2866515
Their 3.5pF capacitance is still sufficient to make them stable, but adding a supplementary 4.7pF can do no harm.
Probably 1N5820 or a similar 3A type.Also checking current capacity of D4, D5, (at least a 5a part) as they do look small but you know I couldn't actually read the labels on the photo.
With 36V supplies, the theoretical peak current is 8.5A on a 4 ohm load, that's 3A average for a sinewave.
But the actual current will be lower, even if there are no parasitic voltage drops: some current is also diverted into the 0.5 and 1 ohm resistors.
Thus, it is safe enough with 3A devices, diodes operated in the forward direction are robust, and supplementary margins are not very useful in this case.
Margins make sense for stressed components, like the OP transistors: any increase in the SOA has a direct impact on the resilience and reliability of an amplifier, and that's typically where efforts have to be focused.
The MJL21194's are an excellent choice in this respect.
Hi Elvee. I'm information hunting again.
D8+D9 as a single zener (D8 zener, D9 short)?
Without running a single 500mw zener harder than 45% current max, how high can its value be?
Without running a single 1.3w zener harder than 45% current max, how high can its value be?
Would a single 5w 1N53XX zener have better audio results than series two of something else?
There was such elaborate care on vas and driver selection, but half of our signal runs through one or two zeners and I think that a zener component selection error may be possible, given that the variety is vast. Is there a type (BZX85xx? BZV85xx? 1N47xx?) of zener more suited or preferable for audio signal or are there (power) types to avoid?
P.S.
Oddly enough, the textbooks claim zener Vf is the same as an ordinary silicon diode; however, most datasheets don't make that claim, since the average zener is 1.2Vf. For BZX Two in series is 2.4Vf, so, this makes a difference fairly quickly. For BZV, Two in series is 2Vf. Does this affect application at D8 and/or D9?
D8+D9 as a single zener (D8 zener, D9 short)?
Without running a single 500mw zener harder than 45% current max, how high can its value be?
Without running a single 1.3w zener harder than 45% current max, how high can its value be?
Would a single 5w 1N53XX zener have better audio results than series two of something else?
There was such elaborate care on vas and driver selection, but half of our signal runs through one or two zeners and I think that a zener component selection error may be possible, given that the variety is vast. Is there a type (BZX85xx? BZV85xx? 1N47xx?) of zener more suited or preferable for audio signal or are there (power) types to avoid?
P.S.
Oddly enough, the textbooks claim zener Vf is the same as an ordinary silicon diode; however, most datasheets don't make that claim, since the average zener is 1.2Vf. For BZX Two in series is 2.4Vf, so, this makes a difference fairly quickly. For BZV, Two in series is 2Vf. Does this affect application at D8 and/or D9?
Last edited:
500mW*0.45=225mW, the current is 13mA, thus 225/13=17VHi Elvee. I'm information hunting again.
D8+D9 as a single zener (D8 zener, D9 short)?
Without running a single 500mw zener harder than 45% current max, how high can its value be?
45V
For a given voltage, two diodes in series will have a marginally higher dynamic resistance than a single, but any effect will be totally marginal, since they are current-driven
No, anything that drops the required amount of voltage does the job, once again because it used in current-mode.
In fact, even a plain and simple resistor works, I have also tested it, the only thing, it should preferably be bypassed by a small capacitor, to avoid enhancing Miller effect in the VAS/splitter
In this case, the Vf is irrelevant since they never become forward-biased.
I was looking at Piersma's board with 35+35vdc rails and then. . . Question: The big Toshiba 2SA1837 driver has a tiny 100ma maximum base current. Is this running within tolerances?
The base current rating has little relevance for linear applications: the important parameter is the collector current, in this case 1A, and the Hfe would need to go below 10 for the Ib rating to be exceeded.
That's impossible in an audio circuit, because it requires deep saturation and a forced current gain of less than 10.
In a class D amplifier, things might be different but here that's no problem.
The 1A Ic rating for the driver is sufficient for all conceivable Circlophone implementations: even if the output transistors have a Hfe of only 20, this allows for 20A output currents, which is more than sufficient.
Limiter testing: For the early warning circuit limiter, a well working zener value is 5.6v. I had to try a variety of tracks to see when the clean clip versus heavy clip region would occur. 5.6v zener sometimes allows a small amount of the clean clip region to be used but never failed to block heavy clipping pulses. With 4n25 (on the limiter schematic), this offers some useful protection. I'm still working on the limiter circuit. . . Miniature Circlophone desktop computer amp, operating at datasheet max, blows 4 ohm speakers right off my desk. lolz!!! Very effective!
I'm also curious on how to measure for Circlophone's bias current with an ordinary digital multimeter. At what points do I measure?
Thanks again!
I'm also curious on how to measure for Circlophone's bias current with an ordinary digital multimeter. At what points do I measure?
Thanks again!
Last edited:
Under quiescent conditions, the (voltage across R8 in mV) = (quiescent current in mA).
You can cross-check with R24/R11: you have to find exactly half the voltage.
If you don't, that means some current is lost somewhere, and it's time to worry!
Across R8, I have 184mv. Earlier I read that the range is 140 to 250. Is that right, and does R21 set this?
That's bang on target, don't touch anything!
Anyway, R21 doesn't set the current: there is no single dedicated component for that purpose, it is the result of balance between current densities in various elements.
Apart from R8, 11, 24, the closest component playing this role is R15, but I advise against tinkering with it, since it also influences the control loop's gain.
Note that the actual value of the quiescent current is unimportant, and has no influence on the quality: it just has to be high enough
That is the Circlophone's philosophy: if the servo has enough room to operate freely, the quiescent current can vary widely without any effect, and increasing it does no harm, but brings no benefit either.
It simply generates more heat.
Getting it right matters more than getting a specific number.
Non-determinism always evil, until Elvee claims it for his own.
Not caring to impose strict order upon things better left smooth,
does not imply that an absurd event such as runaway is allowed.
The universe is safe once again...
---
Fuzzy Logic Crossing, if a meaningful class were to be proposed.
Where we can leave nitpicking details such as "meaning" fuzzy.
"Fuzz'o'logical" crossification works just as well...
Non-determinism always evil, until Elvee claims it for his own.
Not caring to impose strict order upon things better left smooth,
does not imply that an absurd event such as runaway is allowed.
The universe is safe once again...
---
Fuzzy Logic Crossing, if a meaningful class were to be proposed.
Where we can leave nitpicking details such as "meaning" fuzzy.
"Fuzz'o'logical" crossification works just as well...
Last edited:
The current setting mechanism is deterministic, but it is not decided centrally, it is distributed over a number of components: R1 R3 R4 R8 R11 R14 R15 R21 R24 all play some part in setting the current.
The nominal value is set high enough, so that variations due to component tolerances and mismatches always leave enough room for the servo to operate.
This means that in practice, there will be a relatively broad range of standing currents depending on the particular components used, but that is not a problem, and it doesn't make the circuit non-deterministic: it just looks like one.
My first board is up and running and sounds great! It is driving my single driver horns very easily.They are built on Alex's boards.
With reference to Elvee's remarks on bias I duly checked mine. Across R8 I have 214 mA and 87 mA across R11/R24. I am running on a 22v supply although I built for 25v (supply to be built).
The driver transistors need more heatsink as I am not happy with the temperature they run at. Likewise Q5 & Q6.
Any thing to worry about?
It is a lot to ask about, so, I hope that it is okay to ask so much. The case is that not asking is the worst option.
I'd love to see an example schematic of your post#1 Circlophone adjusted for 140ma, and this is for the purposes of comparison. If it is easier just to make a list of resistor values, that's fine too.
Miniature Circlophone has such tiny components that can't shed heat and there's the $18 dual SMPS that can power either heat or speakers but its not especially good for both at once. In a private message that I received (can say either what or who), the sender regarded the components as an insect-like trash. This I found ironic because those components are running our computers fairly well. I'm not saying that we'd necessarily force a Miniature Circlophone to 140ma and "probably" won't put Circlophone into a mint tin either; however, I'm saying that the comparison data could be quite valuable. Surely we could nip off a little heat from several spots and/or enhance the efficiency in some way? Consider the prospects for testing/learning about Circlophone and thus "experimental" schematic is okay for that.
The only clues that I have for it so far, is Odessybmx414's build with 1k5's at R6/R7, one standard resistor value higher than calculated at R21, and really HFE 500+ devices at Q12, Q13 (maybe similar to adjusting R15), but this set of clues is not totally informative.
P.S.
The question I'd actually meant to ask is how to measure for R21 value to confirm that the correct value was/wasn't installed.
P.P.S.
Another area, and especially interesting, is the gain of the drivers. The Harris BD140 that Terranigma loves for audio quality, is actually BD140-16 (HFE 200+), so is it possible to adjust a resistor value to make HFE 150 devices do similar or get closer to HFE 200 results?
Last edited:
D8, D9 is missing from the CFP schematic because working values begin at short. However, coolest values for an amplifier using 25+25vdc are then D8=12v, D9=12v. If using the 1.3w zeners, it is then D8=24v D9=short (a wire jumper), like this.
For reference, see the schematic at post 1 of this thread (especially convenient because it has 25+25vdc operating voltage, just like yours). D8, D9 are for thermal management so their absence could cause too much heat.
Terranigma's latest board shows a few drillings through the board at places where air should be allowed to circulate a bit (instead of hot air trapped at the devices). These anti-heat-pooling measures may be useful to your build as well.
P.S. What is your DC offset, and does it bounce during playback?
Last edited: