and now for something completely different part 3

[audiostar]

1 hour ago, Emooze said:

I am into this. I have been thinking about a CFA3 and an SMD version appeals to me a lot. Hope you can get it working!

Btw, the split boards reach the goal of being able to mount vertically on the heatsinks, so there is no need for SMD 🙂 well, except for the fun of it 🙂 Just kidding.

[mrjayviper]

anybody has a spare PCB (enough for a balanced CFA3 and 1x GRLV)?

Thanks!

[EL_Ken]

I am building a pair of split boards(cfp3smt2splitpre&cfp3smt2splitamp)
It is difficult to find 4.2kOhms&7kOhms resisters while I’m planning to use Vishay/Dale CMF55 series.
And there are only a few choices from Mouser (4.2k:66-RC55LF-D-4.2K / 7k:71-PTF567K0000BYEK)
Should I stay the exact value or use 4.22k&6.98k instead.

Thanks

[jamesmking]

3 hours ago, EL_Ken said:

I am building a pair of split boards(cfp3smt2splitpre&cfp3smt2splitamp)
It is difficult to find 4.2kOhms&7kOhms resisters while I’m planning to use Vishay/Dale CMF55 series.
And there are only a few choices from Mouser (4.2k:66-RC55LF-D-4.2K / 7k:71-PTF567K0000BYEK)
Should I stay the exact value or use 4.22k&6.98k instead.

Thanks

Short answer: in almost all cases high accuracy is not required unless specified otherwise in the bill of materials or schematic. Most builders are using 1% tolerance resistors up to about 1W, except in the most critical of uses, and high wattage resistors are usually 5% or so.

Longer answer:

There are several factors to consider.

A "7K" 1% resistor could actually be between 7.07K and about 6.93K and still be in spec. Similarly 4.2K could be between about 4.24K and 4.16K and still be labelled as 4.2K 1% resistor. So if you are lucky with your 4.22K and 6.98K resistors they could fall within the specs of the 7K and 4.2K anyway.

If you have a good LCR meter or multimeter you could buy multiple resistors, measure and hand select the closest match to the values you want. But this is usually not necessary and requires a highly accurate measuring device.

If you need accuracy without measurement you can go for 0.1% parts but they cost more, typically don't come is as wide variety of wattages and often have lower voltage ratings.

Also consider that all resistors heat up when current passes through them and the heat changes their resistance. The amount the resistance changes depends on the materials and manufacturing of the resistor and is usually specified in ppm per degree of temp rise (lower is better i.e. more stable but also more costly). Unless the use of the resistor in the circuit is critical, around 50 to 200ppm is fine for low wattage resistors and multiple hundred ppm is common for higher wattage... So in circuit, depending upon the power the resistor is dissipating, its resistance will vary from that measured when cold anyway...

It is vital you don't exceed the voltage or wattage rating of a resistor and it is good practice to operate them bellow their ratings for safety and long term reliability. When looking at the voltage rating look for working voltage - this is the maximum voltage it can handle continuously. Some specs don't tell you working voltage but rather the maximum voltage just before failure which is a lot less useful. Some specs don't tell you if the voltage rating is maximum or working... so assume its maximum and the working voltage will be a lot lower. 

Edited August 26, 2022 by jamesmking

[EL_Ken]

3 hours ago, jamesmking said:

Short answer: in almost all cases high accuracy is not required unless specified otherwise in the bill of materials or schematic. Most builders are using 1% tolerance resistors up to about 1W, except in the most critical of uses, and high wattage resistors are usually 5% or so.

........

Thanks for your reply

There are 4.22k0.5% & 6.98k0.1% resisters in my hands.
In worst cases, will be 4.24k&6.97k but still within the 1% range.
So I will use them.

I've already picked up some 1/2W 50ppm 0.1% resisters and keeping the same spec for rest of the resisters.
I think this is probably an overkill. Costly but no regret and happy to have this lesson.

[JoaMat]

14 hours ago, EL_Ken said:

I think this is probably an overkill. Costly but no regret and happy to have this lesson.

Welcome to HeadCase. You are obviously on the right track.

 

It’s a rainy today so I had the time to finish this. Milling of the first board is scheduled at Monday 13:30 local time.

[audiostar]

8 hours ago, JoaMat said:

It’s a rainy today so I had the time to finish this. Milling of the first board is scheduled at Monday 13:30 local time.

JoaMat, looks nice. But don‘t the original boards have 14 heatsinked transistors instead of only 8, what‘s the change here?

[JoaMat]

Only the four output transistors need to be on heat sink. The rest are 1/4W or less. It’s certainly an advantage to have all transistors in triple Darlington on heat sink (I think), but my intention is to only put the four outputs on heat sink and see how that works.

[kevin gilmore]

The vbe multiplier really needs to be on the heatsink otherwise thermal runaway may happen.

[audiostar]

Yes, couple all of them to the exact same temp.

[JoaMat]

Thanks Kevin!

Working on an all heatsink version. Left side is almost ready...

 

[JoaMat]

Seems to work in LTspice. Too simple to be true?

[kevin gilmore]

sure, why not. nelson pass does the same thing with an opto isolator. i've seen various other versions of the same thing over the years, not sure why its not more popular.

[JoaMat]

Thanks, then I’ll give it a try.

[kevin gilmore]

you definitely need another cap to ground on pin 3 of the opamp.

[JoaMat]

Thank you, wilco (will comply).

[JoaMat]

Here is result of today’s work. The board was milled yesterday.

Schematically it’s similar to CFA3. Two of the triple Darlingtons transistors are to-126. All other transistors are sot-23 and sot-26.

To early to say how stable this thing is…

Edited September 6, 2022 by JoaMat

[JoaMat]

Still too early say anything about stability.

Yesterday I built another board. One side worked as supposed but the other side’s bais was stuck at zero whatever I did. When I touched a few places on the board with a test pin the PSU’s current raised so the current limiter kicked in. This happened even with an unconnected test pin. I can build very odd things… that I don't understand.

Bottom line: JoaMat should build himself a Headphone Protector.

[JoaMat]

Issue with bais above solved. Reason was a misaligned 600pF smd capacitor shorted a trace to ground.

When powering the amplifier, it takes some seconds until the bais raises to set value. One side tends to wait much longer than then others. Offsets all channels are close to zero – a few mV at most. No hurry with headphone protection...

[JoaMat]

A working stereo CFA3 something. I’ll wait with chassis until I’m confident it won’t “runaway”. So, now it’s on the floor in our living room. My wife is a very understanding woman.

[swt61]

.

Edited September 8, 2022 by swt61

[JoaMat]

Here is a rebuild Dynahi, now a CFA3 something. CFA3smd engraved on back side of  the Dynahi front panel.

A green LED in the small S.

Diagram of bias (in mA for the first 40 minutes from power on) for servo respective no servo. I think that with the servo it’s possible to build this with just 4 transistors instead of 14 (per channel) on heat sink – time will tell.

[mdr30]

Nice! Heat sinks ready for a T-bone steak? I prefer rare.

[dip16amp]

Here is my CFA2 built on dip 16 sockets as 2 balanced dual mono amps with 2 GRLV power supplies (also built on dip 16 sockets).
Balanced input is from a Benchmark DAC and balanced output is to Grado PS500e with Mogami quad 2534 cable (each amp cabled to each ear).
There are eight sockets with 125 pins used (3 pins are not used on the opamp) and 40 wires. Six wires go the the heatsink components.
First channel is running 150 mA bias and DC offset was 30 mV without the opamp, 0.2 mV with it. Still have to build the other three channels.

 

[Thai Tao]

Hi all,

I'm planning to order some board, do you guy think gold immersion and 2oz copper thickness make different in sound compare to HASL and 1oz?