Fun with Feedback
- IslandPink
- Amstrad Tower of Power
- Posts: 10041
- Joined: Tue May 29, 2007 7:01 pm
- Location: Denbigh, N.Wales
#76 Re: Fun with Feedback
I'm not defending every boutique component, nor suggesting that there's a very clear relationship between cost and performance in the many options out there. We probably agree more or less, eg. that a poor capacitor can spoil a phono stage. It's fun finding the parts that are real value for money. I don't know at what point you consider a part 'boutique' - are the Mundorf ZN boutique ?
ps. this is a great thread, good work again Steve.
ps. this is a great thread, good work again Steve.
"Once you find out ... the Circumstances ; then you can go out"
#77 Re: Fun with Feedback
No I think they are good value. I use them in the mcj3 phono for that reason.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
- Thermionic Idler
- Old Hand
- Posts: 1018
- Joined: Sun Apr 22, 2012 8:22 pm
- Location: Southsea
#78 Re: Fun with Feedback
Happy customer here, they are now the coupling caps in my 300B's thanks to a number of positive recommendations here.
Deck: Garrard 301 - Audio Origami PU7 - NW Analogue DH3S
Phono: Pete Millett LR, Lundahl SUT, AMB σ22 PSU
Linestage: Bruno Putzeys Balanced Pre + Hans Polak relay mod, LCDuino control
Power amps: 300B PP 'Symmetric Reichert'
Speakers: FF225WK Big Mets
Phono: Pete Millett LR, Lundahl SUT, AMB σ22 PSU
Linestage: Bruno Putzeys Balanced Pre + Hans Polak relay mod, LCDuino control
Power amps: 300B PP 'Symmetric Reichert'
Speakers: FF225WK Big Mets
-
- Amstrad Tower of Power
- Posts: 10552
- Joined: Wed May 30, 2007 12:25 am
- Location: South Yorks.
#79 Re: Fun with Feedback
Just finished a three day overhaul and restoration of the amp: all new valve sockets, heater wiring, a mod to the power supply, new ground bus and HT network, and a almost a new set of components in the audio circuit. Coupling caps are the cheapest Clarity Cap SA. They perform very nicely. I’m no longer ashamed of the insides.
Sgt. Baker started talkin’ with a Bullhorn in his hand.
-
- Amstrad Tower of Power
- Posts: 10552
- Joined: Wed May 30, 2007 12:25 am
- Location: South Yorks.
#80 Re: Fun with Feedback
I'm well pleased with the way this has turned out. It is certainly the most successful project I've done to date, both in terms of the sound quality and the amount of engineering and testing know-how gained.
The sound itself is truly special. I would even go so far as to say that it beats my 6B4G mono-blocks (yikes!)
Whether it is due to the new components, chiefly the Claritycap SA polypropylenes, or the fact that the assembly, layout and grounding arrangement is a far better execution I don't know, probably a combination of both, but it has a flow and liquidity to the midrange and treble that I have not heard from any amp I've previously built.
As one would expect, given the feedback technique being used, the bass is clean, firm extended, singularly free from bloat and gets a great performance out of the little Mission speakers. The amp may not be powerful, but has the NVA based solid-state amp easily beaten in the finesse, delicacy, detail and tinkly bits department, if not the sheer welly and authority of the transistor amp. I prefer the balance of virtues coming from the Heretik amp.
I would encourage anyone at my level to give global feedback a go in a single-ended amp. You'll need a lot of open loop gain - that's a non-negotiable requirement. Then you'll need to get to grips with resistor ratios, dBs, poles, zeros, phase, frequency compensation and the interpretation of square wave traces. It's educational and fun. Get it right and the rewards are well worth the effort.
Sgt. Baker started talkin’ with a Bullhorn in his hand.
#81 Re: Fun with Feedback
Hi Steve. I’ve been meaning to try feedback on my current long term project - but I need to get to grips with the theory. What sources (books etc) would you recommend?
-
- Amstrad Tower of Power
- Posts: 10552
- Joined: Wed May 30, 2007 12:25 am
- Location: South Yorks.
#82 Re: Fun with Feedback
Hi Max,
I mainly used RDH4 and Basic Audio by Crowhurst volume 3. Crowhurst was great for the basic feedback theory. Not too difficult to understand.
However, the frequency compensation theory nearly killed me, and I simply could not get my head around the maths involved, so I just used series RC network theory (basically Zobels) instead for the input anodes and parallel RC network theory to get the cap across the feedback resistor right.
Identifying the rogue frequencies involved sweeping a sine wave up with the feedback connected and watching for HF peaks which indicated phase shift, meaning positive or zero feedback at those frequencies, which meant either oscillation or nasty harmonics, neither desirable.
With the scope, I found two HF phase shifts with the output transformer I was using (55KHz and 200KHz) The series network 4K7 and 220pF at the anode gives a 70deg phase shift at 55KHz dropping to 37deg at 200KHz.
The parallel RC circuit formed by the cap across the feedback resistor gives 27deg at 50KHz and 62deg at 200KHz. ( 62-37 for 200KHz = 25deg 70-27 for 55KHz = 43deg) So basically both deal with both frequencies and nothing approaches 180deg.
The combination of these two networks gave a decent settling time for the tops of the square waves at 10KHz. You want no more than two peaks on the square wave trace, but you need to try to keep a decent rise time too, or else the sound will become dull as ditchwater. Fast rise-time and no overshoot and ringing are mutually exclusive with an output transformer in the way. I had three with just the parallel network across the feedback resistor. The series network on the input anodes flattened it to one peak, followed by a slight squiggle (not very technical) and a reasonable rise-time. The improvements were easily heard at each stage of the process.
After a period of despair, I did eventually find a couple of online calculators which helped no end, here:
Series RC - https://www.redcrab-software.com/en/Cal ... es-circuit
Parallel RC - https://www.redcrab-software.com/en/Cal ... el-circuit
The above give loads of detail in the answers including the important phase shift information.
I wouldn't have even had a clue what calculators to look for had I not first read up on the theory, so that was a big positive.
The LF phase shift compensation was taken care of by making making sure that all the resonant frequencies in all the RC networks including the power supplies were below 1Hz, apart from the output transformer one, which I made the single dominant pole.
It's not the perfect way to do it, but at my level of maths it was certainly the most practical. The 'proper' way with Nyquist diagrams and bode plots for the frequency compensation would have been a complete impossibility for a non EE like me.
I mainly used RDH4 and Basic Audio by Crowhurst volume 3. Crowhurst was great for the basic feedback theory. Not too difficult to understand.
However, the frequency compensation theory nearly killed me, and I simply could not get my head around the maths involved, so I just used series RC network theory (basically Zobels) instead for the input anodes and parallel RC network theory to get the cap across the feedback resistor right.
Identifying the rogue frequencies involved sweeping a sine wave up with the feedback connected and watching for HF peaks which indicated phase shift, meaning positive or zero feedback at those frequencies, which meant either oscillation or nasty harmonics, neither desirable.
With the scope, I found two HF phase shifts with the output transformer I was using (55KHz and 200KHz) The series network 4K7 and 220pF at the anode gives a 70deg phase shift at 55KHz dropping to 37deg at 200KHz.
The parallel RC circuit formed by the cap across the feedback resistor gives 27deg at 50KHz and 62deg at 200KHz. ( 62-37 for 200KHz = 25deg 70-27 for 55KHz = 43deg) So basically both deal with both frequencies and nothing approaches 180deg.
The combination of these two networks gave a decent settling time for the tops of the square waves at 10KHz. You want no more than two peaks on the square wave trace, but you need to try to keep a decent rise time too, or else the sound will become dull as ditchwater. Fast rise-time and no overshoot and ringing are mutually exclusive with an output transformer in the way. I had three with just the parallel network across the feedback resistor. The series network on the input anodes flattened it to one peak, followed by a slight squiggle (not very technical) and a reasonable rise-time. The improvements were easily heard at each stage of the process.
After a period of despair, I did eventually find a couple of online calculators which helped no end, here:
Series RC - https://www.redcrab-software.com/en/Cal ... es-circuit
Parallel RC - https://www.redcrab-software.com/en/Cal ... el-circuit
The above give loads of detail in the answers including the important phase shift information.
I wouldn't have even had a clue what calculators to look for had I not first read up on the theory, so that was a big positive.
The LF phase shift compensation was taken care of by making making sure that all the resonant frequencies in all the RC networks including the power supplies were below 1Hz, apart from the output transformer one, which I made the single dominant pole.
It's not the perfect way to do it, but at my level of maths it was certainly the most practical. The 'proper' way with Nyquist diagrams and bode plots for the frequency compensation would have been a complete impossibility for a non EE like me.
Last edited by Cressy Snr on Fri Jul 08, 2022 7:12 am, edited 5 times in total.
Sgt. Baker started talkin’ with a Bullhorn in his hand.
#83 Re: Fun with Feedback
Thanks Steve. I’m certain I have both of those books somewhere so I’ll dig them out. Coupled with your explanation of the process I have a good place to start
-
- Amstrad Tower of Power
- Posts: 10552
- Joined: Wed May 30, 2007 12:25 am
- Location: South Yorks.
#84 Re: Fun with Feedback
Here are the schematics for the audio circuit and the power supply. I've made a few minor changes to the audio circuit since the last schematic, but it is essentially the same thing. The power supply has had a minor change: a 2 x 50uF dual capacitor, rather than the two motley old 47uF and 100uF electrolytics that had decoupled the input and driver stages. The feedback and its associated frequency compensation is matched to the Hammond 125FSE output transformers. Any other make of OPT may require different values. For best results the HF compensation has to be matched to the individual OPT characteristics.
Last edited by Cressy Snr on Fri Jul 08, 2022 4:18 pm, edited 1 time in total.
Sgt. Baker started talkin’ with a Bullhorn in his hand.
#85 Re: Fun with Feedback
Just to be pedantic, you have got the power supply caps and resistors in both diagrams, makes it look like they are doubled up.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
-
- Amstrad Tower of Power
- Posts: 10552
- Joined: Wed May 30, 2007 12:25 am
- Location: South Yorks.
- Dave the bass
- Amstrad Tower of Power
- Posts: 12273
- Joined: Tue May 22, 2007 4:36 pm
- Location: NW Kent, Darn Sarf innit.
#87 Re: Fun with Feedback
Sir, Sir! < question from the not-quite-all-there kid at the back of the class>
What's the function of that 200pF cap in the Anode of the 1st stage Sir? I've not seen that done before. Some sort of eq correction?
What's the function of that 200pF cap in the Anode of the 1st stage Sir? I've not seen that done before. Some sort of eq correction?
"The fat bourgeois and his doppelganger"
#88 Re: Fun with Feedback
Its to reduce the first stage gain at high frequency. If you imagine the 220pf cap starts to look like a decreasing value resistor and as it does the anode load reduces so the gain reduces. Only at AC of course so the DC operating point stays the same. You will see them in a lot of push pull amp designs from in the day. The series resistor is to provide a minimum valve that the load reduces to.
For example:
For example:
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
-
- Amstrad Tower of Power
- Posts: 10552
- Joined: Wed May 30, 2007 12:25 am
- Location: South Yorks.
#89 Re: Fun with Feedback
It’s a step network designed to work at high frequencies as Nick has said above.Dave the bass wrote: ↑Sat Jul 09, 2022 10:07 am Sir, Sir! < question from the not-quite-all-there kid at the back of the class>
What's the function of that 200pF cap in the Anode of the 1st stage Sir? I've not seen that done before. Some sort of eq correction?
Basically, in conjunction with the cap across the feedback resistor it is one part of an attempt to tune the feedback, so that it is equal at all frequencies (not possible in practice but it can be got near enough for it not to be a problem.)
I suppose it’s a bit like the ignition advance on a car engine (itself a form of feedback) which if it wasn’t there would severely limit the high revving capability.
Similarly, negative feedback not properly frequency compensated can and did/does make an amp perform like crap in the treble region, hence the sometimes justifiable hostility to it from many in the audio community.
Sgt. Baker started talkin’ with a Bullhorn in his hand.
- IslandPink
- Amstrad Tower of Power
- Posts: 10041
- Joined: Tue May 29, 2007 7:01 pm
- Location: Denbigh, N.Wales
#90 Re: Fun with Feedback
Interesting ... but why does the response tail off in the treble without this ?
Is there something about the higher open-loop gain followed by feedback, that magnifies the Miller capacitance effects, for example ?
Is there something about the higher open-loop gain followed by feedback, that magnifies the Miller capacitance effects, for example ?
"Once you find out ... the Circumstances ; then you can go out"