PDA

View Full Version : Developing the MA100 amplifier



Radford Revival
12-05-2019, 17:20
Among seemingly a million other projects, I have been developing a new flagship amplifier, using a newly developed topology (for lack of a better word, it pains me slightly to use it as it is so thrown about these days - but I think this just about qualifies)

The first amplifiers to use this "topology" will be 100W KT88 or compatible monoblocs. Somewhat like Radford's own ISA series of amplifiers that were conceptualised but sadly never materialised, the aim is to achieve a genuine step up in performance in valve based amplifiers which I believe I have achieved with this design, and a working proof of concept exists.

The design is inspired by Radford's own TT100 amplifier, in that it is a hybrid design. The output stage is of course valve, but driven via a highly linear differential Class-A transistor stage. The actual amplifier circuit is extremely simple, but afforded very high performance thanks to other innovations, and perhaps to the simplicity itself.

The basic thinking behind this new design is to eliminate or significantly reduce some of the misbehaviours the vast majority of valve amplifiers exhibit - ones that cannot possibly be argued to contribute a pleasant coloration. In doing so, both the subjective and measured performance significantly improves, and the benefits can clearly be demonstrated.

Chasing low distortion was not the goal, but measurements taken from the prototype amplifier, show THD on the onset of clipping at 1Khz, as low as 0.03%. Distortion reduces further with level and at typical listening levels, practically nonexistent. No extreme measures were taken to lower the THD, but this improved linearity is a natural result of removing sources of "misbehaviour" elsewhere in the amplifier.

THD at full power at 10Khz is still as low as 0.2%, which for a Class-AB amplifier using a single pair of output valves, is remarkably low.

This new amplifier will feature a few innovations such as fully automatic biasing (not a new concept in itself), using simple analogue circuitry to set the true quiescent point of the amplifier at all times. This affords a useful advantage in that valve matching becomes far less important for good performance, and eliminates all biasing worries - the bias will be set to a higher degree of precision than is strictly needed, at all times.

There are more innovations to list, so more details, including prototype pictures, to come soon!!

JohnJo
12-05-2019, 17:28
Sounds great Will.

Would be nice if the filament supplies were heavy enough to allow KT120 and KT150 to be used too. Would this require the auto bias set point to be adjusted?

What size of chassis is required roughly?

Firebottle
12-05-2019, 17:49
The actual amplifier circuit is extremely simple, but afforded very high performance thanks to other innovations, and perhaps to the simplicity itself.

Loving the sound of this Will, just the approach I favour, to good result as well. However my amp is valve driver and MosFet output.
Vive la difference.

Radford Revival
12-05-2019, 18:29
Sounds great Will.

Would be nice if the filament supplies were heavy enough to allow KT120 and KT150 to be used too. Would this require the auto bias set point to be adjusted?

What size of chassis is required roughly?

KT120 and 150s will indeed be supported and heater supply specified to cope, and I have tested KT150s. The trade-off with using KT150s appears to be higher peak current delivery at the cost of slightly higher overall distortion. These would work well with particularly difficult loads. The power supply I intend to use will easily cope.

Zero adjustment will be necessary to change between valves, just plug and play. All variances are adjusted for by the bias servo. Zero worries about biasing.



Loving the sound of this Will, just the approach I favour, to good result as well. However my amp is valve driver and MosFet output.
Vive la difference.

Another way to do things! Typically this is what people think of in terms of a hybrid, so I guess these MA100s will be "backwards" hybrids ;)

nickbaba
12-05-2019, 21:11
Very nice, Will. I have an original Radford TT100 and a modernized, upgraded version of that concept will be awesome. Looking fwd!

Radford Revival
13-05-2019, 16:25
I will try to detail some of the philosophy behind this new design in this post.

Valve amplifiers often have fairly poor behaviour near or at full power, and especially for a period after time after even brief clipping. This is responsible for the well known phenomena of valve amplifiers "falling apart" or "turning to mush" into a difficult load or being used near their power limit.

The occasional clipping isn't itself the problem - in fact this isn't even normally noticed. The issue is the long recovery time most valve amplifiers exhibit after the fact. There are actually a few mechanisms responsible, and unfortunately they all interact to make the whole problem even worse. If a valve amplifier is being used near its maximum output power, it is likely these problems will show up.

This new design eliminates these issues allowing much more apparent dynamic range and no "falling apart" even when tasked with driving a horrendous load. The resulting amplifier is solid-state like (to the point where it's uncanny) in its bass grip and damping, yet retains the pleasing low-order harmonic profile of a valve output stage.

struth
13-05-2019, 17:01
Sounds intriguing

JohnJo
13-05-2019, 17:11
Does it use two output valves per channel?

Radford Revival
13-05-2019, 17:17
Does it use two output valves per channel?

Yep - a pair of KT88 or compatible for a target of 100W. The output valves will be pure tetrode connected and not ultra linear (genuine technical reasons behind this) so this power figure shouldn't be too out the way, nor hard on the valves, especially with the features I've not even disclosed yet ;)

Radford Revival
27-05-2019, 19:54
Developments still being made - I also managed to get as low as 0.025% distortion out the thing at full power / 1Khz!

petrat
28-05-2019, 10:19
Will sure does know his stuff .... and he's a perfectionist, so these will be something special, I'm sure.

A year on, and I'm still awestruck by the STA-100 he rebuilt and kindly sold to me. I thought it was my last amplifier, but looks like the new MA-100 will be very, very tempting indeed.

nickbaba
28-05-2019, 13:25
Any thoughts about the form, Will? I imagine form will follow function to a degree, but I know the old TT100 looked very different to the Radford STA series - it's basically a big black metal box!

Will your MA100 look similar to the Radford STA series, or the old TT series, or be something else again?

Radford Revival
28-05-2019, 17:02
Will sure does know his stuff .... and he's a perfectionist, so these will be something special, I'm sure.

A year on, and I'm still awestruck by the STA-100 he rebuilt and kindly sold to me. I thought it was my last amplifier, but looks like the new MA-100 will be very, very tempting indeed.

Great to hear you're still enjoying your amp :D Let's see how much I can temp you with the shiny new ones once they're ready ;)


Any thoughts about the form, Will? I imagine form will follow function to a degree, but I know the old TT100 looked very different to the Radford STA series - it's basically a big black metal box!

Will your MA100 look similar to the Radford STA series, or the old TT series, or be something else again?

The new amp will be visually very much in keeping with the classic older Radfords with handles etc - though slightly rejigged so that it is safe to run without the perforated cage (but it will come with one). They will quite closely resemble the STA100 but be a bit smaller. As monoblocks they won't be so heavy either so far easier to deal with, the STA100 and TT100 are very awkward as a result of their weight - this was one of the main reasons I went with monos.


One feature I have developed worth mentioning is an auto-standby system which will reduce the bias current to around half after a number of minutes with no signal. On the detection of a signal the bias will recover to the full value within a couple of seconds (the transition will be inaudible and seamless)

As many amps spend a signification proportion of their time idle this will have an obvious benefit in terms of valve life and power consumption - a valve amp sitting there reproducing silence clearly doesn't need full idle dissipation. It will be possible to disable this feature with a simple jumper or switch if this does not appeal.

All the advanced features such as the automatic biasing and standby have been kept as simple as possible to avoid overcomplexity - they are all simple analog circuits using very ordinary parts working with no stress or dissipation, they have been carefully specified so that their inclusion is worth the few extra components while not adding needless complication.

There will also be robust protection circuitry to ensure no damage occurs to the amp in the event of a valve failure. No one likes sending things back for repair, or smoked up boards. The aim is for it to be pretty much impossible to blow up even with extremely faulty valves.

CliveG
28-05-2019, 18:09
Like Petrat I would not have thought I could be tempted away from my STA100, but this sounds an attractive proposition. Look forward to hearing more and seeing some pictures!

nickbaba
30-05-2019, 15:36
Sounds nice, Will. Incidentally that would be my preferred look - love the style of the STA series.

Radford Revival
06-06-2019, 19:58
Development has been continuing in the background on this - today I managed to massively increase performance as I had been unintentionally hampering the thing!

thingfish
06-06-2019, 20:08
Will....."You build it, they will come"

Anyway Wot about this new pre-amp ???

Radford Revival
07-06-2019, 15:15
Will....."You build it, they will come"

Anyway Wot about this new pre-amp ???

That's in the pipeline after these monos :D

It has become apparent that they (the amps) will be something rather special.

Essentially this will be the valve amplifier for those who love the idea of one, but also want low distortion and IMD, impeccable overload recovery, tolerance of difficult to unreasonable loads and a wide bandwidth. As far as I can tell this design has indeed pushed the envelope.

southall-1998_mk2
07-06-2019, 19:40
Good stuff, Will. Still lusting after the STA15 Series 5. Maybe someday!

S.

Radford Revival
10-06-2019, 11:20
Good stuff, Will. Still lusting after the STA15 Series 5. Maybe someday!

S.

:D


Tentative rough specifications of this new design, subject to changes as things are tweaked / refined:

Bandwidth 10Hz - 50Khz (100Khz -6dB approx)

THD into 8R / 1Khz:

Just over 1W: 0.004% (Almost entirely just 2nd and 3rd harmonic)

Approx 10W: 0.008% (Almost entirely just 2nd and 3rd harmonic)

Approx 25W: 0.012%

Full power: 0.025% (amount of power slightly curtailed in prototype due to unregulated screen supply dropping significantly)

Yes, this is with a valve output stage, and this is actually significantly better than many solid state amplifiers, even currently sold ones, especially in terms of harmonic spectrum at low levels.

Output impedance approximately 0.3 ohms, substantially resistive even until top of audio band (not directly measured as of yet)

Minimal peaking / ringing into capacitive loads, unconditional stability

Pharos
10-06-2019, 12:28
Very impressive figures Will.

thingfish
10-06-2019, 12:38
Hi Will.

Once this MA-100 beast becomes available, any chance of a trade in........???

Only kidding......The STA 25 MK5 is bloody stunning.

Radford Revival
07-07-2019, 18:48
Pics of the prototype board + 10Khz small signal square waves. Note symmetry, fast rise times and suppressed resonance far above the audio band, somewhere around 150Khz. This could probably be tweaked out but I've not yet tried. If not for the squiggle it is more akin to an amplifier with no OPT at all!

This is with full bandwidth feedback applied.

What will hopefully lead to a very subjectively good sounding amplifier is that the IMD is so low as well as THD. With a 19+20Khz twin-sine test IMD remains around 0.1% - most tone combinations and mixing ratios hover around this level when approaching the power limits of the amp. At lower levels they are proportionally lower.

Recent tweaking has lowered mid-band THD to around 0.02% at full power (even 10Khz THD is 0.03% if you only look at the audio band!)

10W (i.e. where you'd mostly be using the amp) THD is around 0.005%, pretty much entirely just 2nd and 3rd harmonics, and a hundredth the level of many amps. That means 1 / 10,000th the power in distortion components than an amplifier giving 0.5% distortion. By no means state of the art by the standards of some of the latest solid state designs, I'd say this is pretty respectable for a valve amp!

Of course numbers aren't everything, and the circuit doesn't really go to any herculean efforts to get it that low, it is just the result of careful design around known sources of nonlinearity.


https://www.radfordrevival.co.uk/wordpress/wp-content/uploads/2019/07/DSC_1502.NEF_.jpg

https://www.radfordrevival.co.uk/wordpress/wp-content/uploads/2019/07/DSC_1503.NEF_.jpg

https://www.radfordrevival.co.uk/wordpress/wp-content/uploads/2019/07/DSC_1504.NEF_.jpg

Edward
07-07-2019, 19:39
Looking and I'm sure sounding great Will. Impeccable build quality as usual.



Firebottle
08-07-2019, 06:28
..... it is just the result of careful design around known sources of nonlinearity.

Wonderful work Will, little tweaks bring large rewards?

Radford Revival
08-07-2019, 10:20
Wonderful work Will, little tweaks bring large rewards?

Indeed :)

The design is approaching some sort of final form now. Output impedance is now approx 0.2 ohms, which is about where I wanted it, going lower either needs excessive global feedback or positive feedback tricks. I've found around 0.2 seems to be where it gets low enough to no longer matter too much.

An interesting aspect of the design is that this value is maintained throughout the entire audio band - hardly any rise at 20Khz and no rise at infrasonic frequencies.

Many valve amps exhibit a loss of damping and gain peak between somewhere under 10Hz and (obviously not including due to OPT) DC. This is due to an often exhibited LF resonance in most AC coupled designs with feedback.

This might not seem like an issue, however this is right in the region where turntable rumble manifests, and also where many even order intermodulation products end up. If you've ever wondered why your woofers flap around when playing a lower powered valve amp near its limits, even without heavy bass content, this is why. This will then lead to further intermodulation within the speaker itself. Not ideal!

This new design eliminates this issue completely due to the feedback network operating down to DC - there is no subsonic resonance or loss of damping, so solid state-like damping is available!

Radford Revival
16-11-2019, 20:34
The silence on this front is no indication of lack of activity!

Between other jobs I have been optimising the design further. I believe I may have found another degree of freedom with the design to further increase performance with no additional complexity...

Case work is also now finally in development and the chassis as been laid out. This will be worth the wait.

nickbaba
30-01-2020, 14:15
Any further news on this, Will? Might 2020 be the year??

Radford Revival
31-01-2020, 20:49
Any further news on this, Will? Might 2020 be the year??

Hi Nick

That's indeed the intention - it is largely a game of Tetris getting the thing into the chassis now. I know exactly the general layout of the thing, I just need to complete PCB designs for the power supplies, which thankfully I've managed to simplify the design of somewhat. Another nicety of the simplification is that all the rectification and "noisy" side of things will end up in its own compartment shielded from the rest of the amp.

There is a prototype pair being built on a set of off the shelf chassis which are close to the final dimensions. I will admit to getting sidetracked for a bit on a few side-issues but I've reigned it in now. Thankfully going off on a tangent paid off and I have managed to improve things in key areas.

There are a couple of puzzles left to work out but the actual concept is fully sound and proven. As the first pair come together I will be sure to post pictures here.

Radford Revival
14-12-2020, 22:38
Just a small update - this project is most certainly not dead!

Despite being somewhat unwell for much of this year I have been plugging away in the background on this and have actually improved upon the design massively - it has turned into something rather special. I am actually somewhat glad things got delayed - I've managed to simplify the design massively AND improve performance at the same time! I had also identified some subtle design flaws in the original concept which have been rectified. This is why I'm glad I didn't plough forward with some earlier version. There have been over 50 versions of this circuit but they are rapidly converging on a pleasingly simple solution....


The final amplifier will have performance previously thought impossible from a push-pull transformer coupled valve topology.

Firebottle
15-12-2020, 09:05
- I've managed to simplify the design massively AND improve performance at the same time!

That's fantastic Will, that's always a win-win.
Hope your health is improved now.

Seasons greetings.

Radford Revival
15-12-2020, 13:16
That's fantastic Will, that's always a win-win.
Hope your health is improved now.

Seasons greetings.

Thanks Alan - I'm getting there. It has been a tough year. Hope you have a great Christmas!

Marco
15-12-2020, 15:29
Just a small update - this project is most certainly not dead!

Despite being somewhat unwell for much of this year I have been plugging away in the background on this and have actually improved upon the design massively - it has turned into something rather special. I am actually somewhat glad things got delayed - I've managed to simplify the design massively AND improve performance at the same time! I had also identified some subtle design flaws in the original concept which have been rectified. This is why I'm glad I didn't plough forward with some earlier version. There have been over 50 versions of this circuit but they are rapidly converging on a pleasingly simple solution....


The final amplifier will have performance previously thought impossible from a push-pull transformer coupled valve topology.

Fantastic stuff, Will. I've no doubt that the final build will indeed be rather special. All power to you, my friend. You're a credit to AoS and today's electrical engineers. Hope you recover fully, and as soon as possible, from your illness.

Have a lovely Christmas:christmas:

Marco.

Radford Revival
15-12-2020, 16:19
Fantastic stuff, Will. I've no doubt that the final build will indeed be rather special. All power to you, my friend. You're a credit to AoS and today's electrical engineers. Hope you recover fully, and as soon as possible, from your illness.

Have a lovely Christmas:christmas:

Marco.

Thanks for the kind words Marco :) Hope you also have a great Christmas!

Radford Revival
22-02-2021, 13:46
Work is ongoing - many design refinements have been made - especially to the drive circuitry. Each output valve is now driven by a 1Mhz bandwidth ultra-low distortion class A driver circuit - the wide bandwidth for technical reasons, not bragging rights.

The bias servo has also benefited from some refinements and simplifications, and the chassis design is also finally taking shape.

Edward
22-02-2021, 14:23
Very much looking forward to seeing some pictures of the MA100 Will. Sounds like you are making steady and important progress with this project.

Chat soon. Please say "Hi" to your Dad. Hope you both keeping well.

Edward

Radford Revival
25-02-2021, 12:47
Very much looking forward to seeing some pictures of the MA100 Will. Sounds like you are making steady and important progress with this project.

Chat soon. Please say "Hi" to your Dad. Hope you both keeping well.

Edward

Hi Edward

Hope to post some pictures when the prototype chassis is up and running - we have some off the shelf chassis that are roughly the right size to test all the PCB assemblies before moving over to custom metalwork. :)