DIY Front End 2022

Jump to Latest
#261
Driving a transistor to immediate failure and using it for long term reliability are two different things.
And any industry would use safety factors, only how safe is safe.

I certainly would not want my electronics to fail.
My speakers are too expensive, and I am not as rich as Nelson. 😉


Patrick
 
#263
I was delighted to see the attached figure, in Nelson's technical paper about this circuit. It's one of those thrilling situations (to me anyway) in which academic theory and practical reality agree beautifully. Just exactly as James Solomon predicted in 1974. diyAudio member @Bonsai kindly makes Solomon's tutorial available online, here .

In a different thread, Nelson had this to say:

Nelson Pass said:
... The J113 as an example has lower transconductance than the K170 Jfet types, and so will have less capacity for processing feedback. ... This issue can be addressed by using different quantities of them in parallel and degenerated, but it's a pain in the butt. Easier to spend a little money on the K170s.
Click to expand...
_
 

Attachments

  • distortion_and_gm.png
    distortion_and_gm.png
    21.8 KB · Views: 277
  • Like
Reactions: PierreQuiRoule, grataku and pfdavid
#264
#266
Slightly off-topic question - I've been thinking about a DIY integrated amp with an input/buffer stage and maybe an F4 or F5 output stage. Would this front match with an F4 or F5 - with a regulated PS for the front end the universal PS for the output?
 
#267
Aleph5 said:
If you add a duplicate of R6 to the other drain leg of the diff/quad pair, I don't see an excursion problem with them being driven by ±32V. All of the other transistors not so, but there are other options there.
Click to expand...
The voltage on the drain(s) attached to R6 do not get below about 1 volt less than the V+ supply. Thus Vds(Q1) will be about 2 volt less than the V+ supply input. The variation of Vds will approximately 1/10 the variation of the output. As a result, Q1 and Q2 should not see a Vds above about 35V, assuming 32V supply to the FE.
 
  • Like
Reactions: Aleph5
#268
bmdduck said:
Slightly off-topic question - I've been thinking about a DIY integrated amp with an input/buffer stage and maybe an F4 or F5 output stage. Would this front match with an F4 or F5 - with a regulated PS for the front end the universal PS for the output?
Click to expand...
Well the F4 is a big power buffer "hot follower" with 0db of gain but lots of current drive capability. All of your voltage swing needs to come from the FE / preamp / gainstage whatever in front. The F5 has ~15db of gain. So, maybe a preamp with a gain of 2 is suitable for that one? It's important to figure out what kind of gain you really need for your own ears, room, listening habits, speakers, etc. when you're trying to connect all these "LEGOs for adults". And the amp / speaker interface matters. For example, certain amps just don't really excel with 4 ohm speakers. The FE presented here reminds me of the old original BA-1/BA-2 FE, another simple discrete op-amp. Similar but different. Complementary BA-2 output stage is just like the F4. But F4 has that little buffer up front that may or may not be needed depending of what people are hooking to these things. The single-ended BA-1 OS is really nice, lived with that one for awhile. There are pros and cons to all this stuff. So, the answer is very simple..... keep reading and learning, build everything, and enjoy the journey.... 🙂
 
#270
lhquam said:
The voltage on the drain(s) attached to R6 do not get below about 1 volt less than the V+ supply. Thus Vds(Q1) will be about 2 volt less than the V+ supply input. The variation of Vds will approximately 1/10 the variation of the output. As a result, Q1 and Q2 should not see a Vds above about 35V, assuming 32V supply to the FE.
Click to expand...
True, but there would be a little more margin for signal excursions with another R and it wouldn't hurt.
 
#272
I have changed R6 to 680 Ohm, as in the original schematic. I had to adjust R7 to 660 Ohm, to get a very low output-offset (below 1mV).
I measured only 1.2 V over the 660 Ohm resistor (R7) = circa 2mA over the JFets - hmmm 🤔
seems low for me? Sounds good.
In my first version I had R7 at 330 Ohm and R6 adjusted to circa 50 Ohm. In the measurement this was slightly 3rd harm.dist. dominant.
In my second version -with resistors as mentioned above- the measurement showed more 2nd harmonic distortion.
Only some of my experiences - I don't know if they have any value?
Cheers
Dirk
 

Attachments

  • Frontend_2022_R6_680Ohm_R7_660Ohm.jpg
    Frontend_2022_R6_680Ohm_R7_660Ohm.jpg
    167.6 KB · Views: 233
  • REW_Frontend2022_measurement_in2V_10kloadout.jpg
    REW_Frontend2022_measurement_in2V_10kloadout.jpg
    95 KB · Views: 230
  • REW_FrontEnd2022_2Vin_10kO_load-measurement2.jpg
    REW_FrontEnd2022_2Vin_10kO_load-measurement2.jpg
    95.6 KB · Views: 237
#276
I am just guessing. The certain offset is needed to get the best performance.

R6 is 680, assuming 0.6V Vbe for Q5, the current thru R6 is 0.9mA. Mr.Pass mentioned slightly <3mA for the first stage. Q2 should have around 2mA. Thus there should be a positive offset at output.

If we do the FE in very traditional way, the current thru Q1 and Q2 should be roughly equal to minimize the offset.

That is why I tried to reduce the current of the first stage to <1.8mA.

But people with inside knowledge never said to minimize the offset to get the best performance.

So far we don't know the offset of the reference schematics. I wish 6L6 could give us some hints.

Since there is capacitor at output, do we really care to minimize the offset?
 
#277
@6L6, @william2001- thanks for your replies. My speakers aren't really demanding 4 ohm loads - I use a pair of Zu and a pair of Linn Ninkas.

I don’t play very loud. The F4 seems to be a great sounding amplifier but requires some gain up front. I think the gain on the new front end is up to 10db?

Some amplifiers don’t need more than a pot to control volume - maybe the F5 is in the category. Also wonder is this front end with the F5 is too much rain overall for my use.

Thanks again - bmdduck
 
#279
wwwtttwww said:
I am just guessing. The certain offset is needed to get the best performance.

R6 is 680, assuming 0.6V Vbe for Q5, the current thru R6 is 0.9mA. Mr.Pass mentioned slightly <3mA for the first stage. Q2 should have around 2mA. Thus there should be a positive offset at output.

If we do the FE in very traditional way, the current thru Q1 and Q2 should be roughly equal to minimize the offset.

That is why I tried to reduce the current of the first stage to <1.8mA.

But people with inside knowledge never said to minimize the offset to get the best performance.

So far we don't know the offset of the reference schematics. I wish 6L6 could give us some hints.

Since there is capacitor at output, do we really care to minimize the offset?
Click to expand...

fact - some ppl are taking one Vbe as 600mV, some as 650mV, some as 700mV

ignoring that really depends of exact part and current trough, it is basic figure used as guide; tight guide, but not exact number carved in stone
 
#280
lhquam said:
Putting a resistor on the drain(s) of Q2 would have little effect on the resulting voltages and distortion. Run the simulation.
Click to expand...
That's what I'm talkin' about. I did run it and of course it has no detrimental effect, but many would say just build it. For me, all in good time...

My real point was that the J113s are not limiting the rails, except for the current sources, but those don't really need to be J113s.
 
You must log in or register to reply here.
Top Bottom