current feedback electrostatic amp

[Laowei]

Thanks Jose! 

[2b青年爱发烧]

I think it's hard to obtain good performance because usually the CFA needs a proper value of feedback resistor, but the resistance in electrostatic amps is too large.

[jose]

My CFA is already singing. Thanks to Kevin, Congo5 and Michael

I´m using a SE source without problems (bridge "cold" channel to ground). My unicial assembly is a bit precarious but only because I wanted to check that everything works correctly.

My "Mini BH PSU" to -407Vdc and + 404Vdc. Offset at the start with 14.5/9.22Vdc channel L and 13.76/11.88Vdc channel R. After one hour it is very similar on all channels. Balance 4Vdc and 2Vdc. I hope this is not worrisome.

I am using "adequate" heatsinks ?. Temperature transistors after one hour of use is barely 40º. I guess that's why the offset remains so stable.

 

Edited March 12, 2019 by jose

[Laowei]

I’m almost there too.

Did you use 2K ohm for your base resistors, or another value? 

 

[jose]

I have not changed any value. I guess you mean R33 / 35 according to Dr. Gilmore's scheme ? Yes, 2K. 

 

[Laowei]

Yes, those are the ones. Joamat had experimented with different values and was wondering if you did too. Got to order 4 more for the other channel.

As usual, a great looking build, Jose.?

[jose]

26 minutes ago, Laowei said:

Yes, those are the ones. Joamat had experimented with different values and was wondering if you did too. Got to order 4 more for the other channel.

As usual, a great looking build, Jose.?

Thanks but my amps are very simple... besides this is a temporary wiring. It's all pretty chaotic. 

For the moment, I prefer to start with the values that Congo5 selected. Anyway, with these heatsinks, my amplifier stays moderately hot, so I rule out changing anything if it's not necessary.

Edited March 12, 2019 by jose
orthography

[congo5]

1 hour ago, jose said:

the values that Congo5 selected.

my board is a direct copy of Dr Gilmore's layout and values.

so all credit goes to him.

I saw no need to make resistor value changes.

Some things can be adjusted as Michael did but with enough heat sink it works just fine as is.

A stable temp is good if you want stable current through the output stage

think mine were about 18ma warmed up on 400v rails

I have two working amps and no plans to mess with them.

Thanks

[JoaMat]

As DIYer you are entitled to do whatever you like. IMO 18 mA is at the higher end. For the moment I run mine at one fourth of 18 mA. I’ve my house full of heat sinks but that’s not a reason for high currents.

[congo5]

so 4.5ma?

I did not change it cause did not feel I needed to

would think 9ma is enough

not telling people what to do

just what the stock values were from Kevin

 

[Laowei]

I was only curious what Jose was using, as my heat sinks are about the same size as his. For first power up, I wanted to be able to run them awhile without worry as a baseline.

Did not wish to cause big confusion here for no reason. Thanks to each for your input.

[congo5]

looks like I'm the one that caused trouble, not you....

those sinks above are much bigger than mine.

you should be fine but do watch temps and measure across the 100r if you want to know what its running at ...

[jose]

My amplifier was sounding about 5 hours continuously. I had no problem handling 18mA. Temperature with this heatsinks makes all measurements very stable.

The body of the transistors was at about 40º after one hour and it remained in a range of about 40 ~ 45º all the time.

I think it's right in choosing big heatsinks.

As I mentioned, working with SE sources the amplifier goes perfectly, although with a balanced source the volume increases a little.

I do not see anything wrong with modifying the BIAS. Maybe a trimpot could be added to future pcb´s ¿?. I will leave it at 18mA for the moment.

What worries me a little is the offset. Although it´s stable (hardly varies ~ 0.7Vdc after being so long on) I don´t feel very comfortable with values of 14.5 / 9.22Vdc channel L and 13.76 / 11.88Vdc channel R maybe because I'm used to having levels close to 0Vdc. 

[mwl168]

5 hours ago, jose said:

What worries me a little is the offset. Although it´s stable (hardly varies ~ 0.7Vdc after being so long on) I don´t feel very comfortable with values of 14.5 / 9.22Vdc channel L and 13.76 / 11.88Vdc channel R maybe because I'm used to having levels close to 0Vdc. 

That's about the offset values I am seeing in my build too. I don't think it's a problem to worry about since they are below 15V.

I consulted with Kevin a while ago about possibility of bringing down the offset and he think it would involve resistor trimming...  

[mwl168]

On 3/13/2019 at 3:26 AM, jose said:

What worries me a little is the offset. Although it´s stable (hardly varies ~ 0.7Vdc after being so long on) I don´t feel very comfortable with values of 14.5 / 9.22Vdc channel L and 13.76 / 11.88Vdc channel R maybe because I'm used to having levels close to 0Vdc. 

Have been having some offline discussions with JoaMat and Jose regarding the possibility of trimming the offset if desired. JoaMat came up with the idea of trimming the 100R resistors in series between the KSC2752 emitters and B-. He modeled it in a LT Spice model and I decided to give it a try.

With my amp running at 14mA bias current, without this mod all the DC offset was between + 9VDC to + 4 VDC and balance was under 6VDC and I consider it totally fine. This exercise was more experimental and to satisfy the OCD part of me. 

I put a 50R resistor and 100R trimmer in series and adjust the sum to 90R based on JoMat's calculation. It can certainly be done with a single 150R trimmer,  a fixed resistor in series with a trimmer is to protect against the small odd should a trimmer fails.  I stopped making further adjustment once I got all the offset within 0.5VDC and balance within 0.2VDC. 

A few things to observe:

1. Turning one phase DC offset lower raises the offset in the opposite phase slightly. This gets a little tricky trying to zero out offset of both phases at the same time.

2. The bias current raises slightly as I adjust down the offset. In my case, the resulting increase in bias current was about 0.7mA. This matches what JoaMat saw in his LT Spice model.

3. The final resistance on all 4 "combo trimmers" fall between 86R - 90R. Take this as a reference only as this will vary from amp to amp and dependent on the bias current setting.

I listened to the amp both before and after adjustment and frankly I could not detect a difference.   

 

Edited March 18, 2019 by mwl168

[jose]

Nice Michael !!!!

I am going to try the same Mod but changing 100r for 120r and add a trimmer of 1K in parallel.

In theory the result should be "similar" but who knows. Without magic smoke, not can be glory ... I suppose 

 

 

[jose]

Different 121r and 1k trimmer in parallel.

Ok, it's no nice but we haven't space on the PCB.

[jose]

Offset around +-150mVdc and balance at 100mV in both channels. The resistive value (121R + 1K trimmer in parallel) after regulation is 79.9 Ohm in all channels. My amp passed from 18mA to 20mA BIAS.

This works and  I'm quite plaised. I could leave it like that but I'm thinking about how to do this mod something more "elegant" and at the same time I with this kind of trimmers but I dont´want to separate the pcb to heatsink because is a pain.

 

[JoaMat]

Shorted out the 150K resistors in red circles which then bypasses corresponding Darlington pairs.

I run amplifier +/-400 V and 5 mA. IMO the amplifier sounds about the same as with higher currents and all transistors on duty. I believe it’s possible to build a decent all SMD version of the CFAE. Reduce voltage to let say +/-300 V - how small can one build a power supply?

Edited March 20, 2019 by JoaMat

[Kerry]

I like small.  Sounds like some fun to be had  

[spritzer]

We could do a simple switch mode unit... 

[Laowei]

Another Gilmore CFAE is first powered up. No smoke, no POP,  no drama. 

I used the first GB split PCBs for this build. All part values per the PCB.

Powersupply for the moment is a BHSE from an old KGST: B+ = +353.8;  B- = -358.1

Left Channel    Offset: +7.19/+10.53;    Balance: 3.34;    Bias(mA): 8.28/8.24/8.23/8.15

Right Channel   Offset: +8.42/+8.91;     Balance: 0.49,   Bias (mA): 8.35/8.23/8.68/8.60

After an hour, heat sinks are barely warm to the touch.

Hooked up with an iPhone X direct input as source and STAX 207 it sounds very promising.

Will swap in a TKD volume pot soon to see if things improve ditching the distal volume control. 

Thanks to all, especially Kevin, Congo5, Joachim, Michael, and Jose.

 

 

 

Edited March 25, 2019 by Laowei

[joehpj]

Chassis just arrived. Looking good.

Edited March 26, 2019 by joehpj

[jose]

I think mine is already finished but I need to touch up something.

My pocket CFA. 

 

 

 

 

[Laowei]

Really nice build, Jose.