♫♪ My little cheap Circlophone© ♫♪

[Elvee]

Originally Posted by danielwritesbac

I'm still very interested to replace the Zeners with a transistor, except not the same as last time that made a kink in the treble as if the cap parallel to the feedback resistor quit working (weird and painful ear pressure--high treble went off the beam--feedback didn't work right).

Problem is, I hear no difference (OK, I have tin ears), and more worryingly the effect of the transistor is undetectable by measurements, and even more worryingly, the sim is unable to show any difference.
Without knowing exactly what effect is a problem for you, it is difficult to combat.

This time, I don't want to plug that transistor into the feedback loop. Is there a way?

If you want to mimic more closely the effect of zener with a transistor, you can tie its base to the ground instead of the output.

This configuration is inferior in principle, but shows no practical difference.

[terranigma]

I'm about to buy some BDY73's for my next Tringlophone project. It is more like a high Hfe version of 2n3055 (Vce=60V). Not a problem for single 32V rail. Datasheet graphs shows that It handles 70V if Rbe=1-100 ohms.

Since that Circlophone contains 15R at their 2N3055's B-E legs, can we assume that BDY73's suitable for using in Circlophone with 32V+32V rails? I just need some approval.

[Elvee]

Originally Posted by terranigma

Since that Circlophone contains 15R at their 2N3055's B-E legs, can we assume that BDY73's suitable for using in Circlophone with 32V+32V rails?

In principle, no: the SOA only extends to 60V. The Vcer applies only when the transistor is off.
That said, they will certainly not blow into your face for a mere 4V excess....

[Elvee]

This message is for Terranigma: he has tried a darlington version, and has encountered some overshoot issues with a squarewave signal.

The problem is not very apparent in sim, anyway here is a logical path to a fix: increase C6 to 82pF for example (or any suitable value).

To avoid creating other problems, it could be wise to include a series resistor with the additional cap.
Here is the modified schematic (addition of C4/R2).

The value of R26 could also be tweaked (increased)

Adding a suitable input lowpass could also help (Fc~=200KHz)

[Piersma]

In addition to Elvee's comments on the overshoot issues:

Don't set the input filter to high this woud also add to the overshoot problem(s) especially when it comes to high slew square signals.

[Piersma]

Originally Posted by Piersma

In addition to Elvee's comments on the overshoot issues:

Don't set the input filter to high this woud also add to the overshoot problem(s) especially when it comes to high slew square signals.

Don't set the bandwidth of the input filter to high....

[terranigma]

Originally Posted by Elvee

Here is the modified schematic (addition of C4/R2).

The value of R26 could also be tweaked (increased)

Adding a suitable input lowpass could also help (Fc~=200KHz)

Adding a ~1.5 nF input filter cap completely took care of that overshoot issue for me. But this value is effective while using a 10K volume pot before amp. 2K5 pot requires ~3.9 nF for clean square waves. I don't know why. 10K seems to me not that out of spec at my setup. I used ~550 Hfe BC550C's as input transistors.

Higher R26 values and first attempts for adding extra network at feedback caused oscillations. I didn't try many combinations there and I settled with input filter method.

Thank you Elvee and Piersma.

[terranigma]

Hello All,

I consider that I finished my PCB layout of N-Darlington output version of vanilla Circlophone. I built two channels and they are running for a while. Purity, dynamics, punchy bass.. I think this version is no less than original output scheme. Even with those ST TIP142's. During first tests, I attached Philips BDV65B's and accidents costed 2 pairs of BDV's to me. Then I decided to use cheaply available ST TIP142's and I saw that nothing wrong with these ones. They perform really good and they feature much robust casing.

I used BD140's as servo transistors because mine were good quality ones from Philips and Harris. I measured their cb capacitance at 0V and Harris ones were always lower (35pf vs 52pf). Then decided to to use them although reputability of PH ones.


Edit... additional info from terranigma is here,
Building Elvee's Circlophone: Documentation, Parts, Accessories, & beginner friendly

[keantoken]

Don't just measure Ccb! Measure Cbe too. There are many low-Ccb transistors that have excessive Cbe. Never heard of Harris, BTW. Where do you get those?

[terranigma]

Originally Posted by keantoken

Don't just measure Ccb! Measure Cbe too. There are many low-Ccb transistors that have excessive Cbe. Never heard of Harris, BTW. Where do you get those?

What at least I know about Harris is, they acquired RCA semiconductor. I think they were in production during 90's. I found those BD's from a local shop in NOS condition.

Ok, I did a quick test for Cbe:

Ph BD140 : 120pf
Harris BD140 : 150pf
Moto 2N3020(3019) : 65pf
Moto 2N2905A : 35pf
Ph 2N2905A : 30pf

Toshiba 2sc5171 : 645pF (16pF Cob spec'ed)

So, what this mean? Is Cbe a critical value for bias servo here? Do we consider that BD140 ones obviously out of spec?