ZinAmp at Nov 9 th and 10 th 2019 San Francisco, CA ZinAmp .. in a - - PowerPoint PPT Presentation

ZinAmp

at Nov 9th and 10th 2019 – San Francisco, CA

ZinAmp .. in a Nutshell

• Hybrid Tube / Lateral-FET Integrated
• Tube Phono and Pre-amps, ‘loosely’ based on Marantz 7
• Power Amp – Class A/B Lateral FETs with Class A BJTs - in parallel
• Inspired by the GEM amp
• GEM created by Graham Maynard
• Output for additional power amp
• Because you never know…
• Auto-power off..
• ..and on again .. even with a vinyl record!
• weird, but fun to demonstrate 
• EQ Toggle Switch
• Flat and Dynamic settings
• Similar to a loudness curve, but quite subtle/gentle
• 120W Per Channel – probably!

Lid off !

..and on again!

Block Diagram

Why Bother?

• Modern mainstream amps in 2016 sounded underwhelming to me
• Got interested in vintage equipment – 70s & 80s Technics et al
• Research led me to a “dream combo” of a Hafler DH200 and a Marantz 7
• Couldn’t find these affordably in the UK
• Wanted a Tube Phono Integrated with a Class A output – didn’t exist!
• I was going to have to build this amp!
• Found a really good mentor to help me

My Mentor

Henry Dulat – UK “Top Repairman”!

Acknowledgements

Doug Self Rod Elliot - ESP https://sound-au.com

In memory of…

Graham Maynard

• UK DIY Audio and Radio enthusiast
• Creator of the GEM Amp
• Respected contributor to diyaudio.com
• Passed away in October 2014
• www.bovan.net

• Conventional design with …
• Class A/B stage
• Single Ended Class A stage
• Two local feedback loops
• Quiescent supply of 600mA
• Carlos from Brazil built one

The GEM Amp

Graham’s Mission

• To achieve the sound of John Lindsley-Hood’s 1969 Class A design
• 100-200W power – needs Class A/B
• Crossover distortion in A/B made worse by back EMF from loudspeakers
• Not visible when testing with steady sine waves into resistors!
• but is present with real music signals into speakers, due to transient back-EMF
• Parallel Class A stage mitigates effects of back EMF
• Smoothing the crossover region
• Similar idea to Quad 405 current-dumping design from 1978
• More info at www.bovan.net

John Linsley-Hood - 1969

My Challenge

• To get the GEM amp to work in my ZinAmp integrated box
• I have respectable heatsinks, but not like mono-blocks!
• Limited space for filter capacitors
• Avoiding noise and hum is a key constraint…
• ….followed by space inside the box

Keep it Simple (Stupid)

Design Decisions

• Push-pull Class A would be more efficient
• Suggested by my mentor, Henry Dulat
• Should requires less quiescent current
• Cooler running than the GEM single-ended
• Lateral Mosfet A/B Stage
• Interest in Hafler led me down this path .. prototypes already developed
• Requires fewer components than BJTs
• More thermally stable etc.
• Regulated Power Supply
• Smaller capacitors but a little more dissipated heat
• Remove one of Graham’s LFB loops
• Feeding back into –ve rail
• John Broskie (tubecad.com) suggested this would be noisy at high-volumes
• Broskie also suggested a regulated supply

Power Amp - schematic

• Push/Pull Class A BJT Output
• in parallel with
• FET Class A/B Output
• Local Feedback to VAS
• Bootstrapping for VAS
• Additional Bootstrap driving +ve side
• f Class A stage

Driving the Push-pull Class A

• VAS with parallel followers
• VAS Follower 1
• VAS Follower 2
• Collector of each follower drives
• ne side of output
• Both followers are bootstrapped

Power Amp - PCB

Regulated Supply - schematic

• Conventional Design
• Rod Elliot / ESP
• +/- Rails
• Load Limiter
• Big Power Transistors
• Darlingtons
• 10-15A rating
• FB transistors also need to be fairly

robust

• Dissipates a little heat
• Does the job!

Regulated Supply – PCB

Setup and Biasing

• Aim: to maximise the amount of class A current without overheating!
• Using double-die FETs so twice the A/B bias current
• Class A stage will not ”mop-up” c/o distortion from under-biased FETS
• FET bias current of < 70mA can sound ’chalky’ (Exicon FETS)
• Sweet Spot
• Class AB biased to 75mA – or double-die FETs to 150mA
• Class A biased to 200mA
• Sounds nice and doesn’t run too hot
• Class A stage generates most of the heat though
• Tempted to try mono-blocks with bigger heat-sinks!!
• Plenty of scope for experimentation … and a bit more current!

Setup and Biasing cont..

• Voltage across Rbias = 700mV
• VGS = 350mV per device
• underbiased
• Voltage across Rbias = 900mV
• VGS = 450mV per device
• Sweet spot - sufficient
• Voltage across Rbias = 1200mV
• VGS = 600mV per device
• Ok with single-die FET
• Double-die FETs run hotter

Trade-offs

• Rails can’t go beyond 48v without encountering heat issues
• Tried 58v - sweaty – approx 160° F
• Power output intrinsically limited to about 120W pc music-power
• How much power do you need?
• Push-pull Class-A vs Graham’s Single Ended
• Better Rail Rejection – although regulated supply mitigates this
• Lower current, less heat
• Lower second order harmonics perhaps no longer predominant as with single-ended – although

class A stage is only providing 1 watt or so.

• This is a work in progress!
• Subjective listening is good….
• …though full measurement/analysis is required
• Bob Cordell measured a THD of approx 0.07% yesterday, mostly lower 2nd 😏

How does it sound?

• Really good … well, I would say that!
• Definite step forward from the Class A/B amp I had built previously
• I wouldn’t go back
• Gives a sense of space between notes and sounds – no smear
• Like the blacks in the picture on a good plasma-screen TV
• Vocals and drums sound particularly life-like
• Anything that’s acoustically mic’d
• My wife likes this amp – and she has better hearing than I do👃

Please come and listen

Questions welcome – thanks!

Why not just over-bias the FETS?

Over-bias FETs into Class A:

• Moderate heat dissipation
• Class A at moderate volume

| At the expense of |

• Slight loss of headroom
• Loss of Class A performance at higher volume

Class A/B in parallel with Class A:

• Moderate heat dissipation
• Class A at moderate volume
• Retaining headroom

| At the expense of |

• Loss of Class A performance at higher volume

Either way, as you crank up, the speaker back-EMF increases and the Class-A effect diminishes

| Although with the parallel class A stage, you get a little more headroom |

…but the GEM amp was simply too interesting not to try!