Saturday 21 October 2023

Audio Meet - Owston 21st October 2023

This was my first Owston since 2019 and it was great to meet up again with old friends. It's been an odd few years on a number of fronts. The pub had now changed to Dario's Ristorante, on old restaurant name in Doncaster from a good few years ago.

The weather over recent days had been very wet thanks to Storm Babet and the local watercourses were as high as I'd seen. Fortunately play got underway shortly after the usual 11am start, though the outfield was rather soggy...

(Some of the photos aren't focused very well. I'd forgotten how difficult it is for my camera to focus in the low light even with flash assistance. A slightly blurry photo is better then no photo though,)

First Steve gave his customary welcome...

...and then a team photo for absent friends.

First to go was Steve (no. 2)

Steve was using his recently built speakers (discussed here from around post 169). The system comprised an Innuos ZEN music server, a Nadja preamp with some DSP, a 6 channel power amp. The speakers have a Beyma TP-150 AMT up top with horn built in to the speaker cabinet, a Beyma 12P80MD midrange (running between 70Hz and 1300Hz with a little DSP help), and two Seas Elroy subwoofers.

I think these were Steve's too, though I didn't hear them

Next was Chris who had put the speakers together that morning at home, pretty impressive really.

The amp was a Powersoft digital amp which has DSP, the source was a laptop running Dante (used in PA work a lot apparently)

The subs were Chris' usual PA-duty speakers - Faital Pro 18XL1800 (I think) in a sealed cabinet, The compression driver behind the horn was a BMS 4592. The horn itself Chris had got cheap and he wasn't too impressed with it. He found lagging the outside with bitumen type car damping panels and the throat with a bit of polyfil tidied things up somewhat.

Mike followed Chris with his EL84 PP amp and old speakers from his dad's old music centre with a replacement 4" full range driver and tweeter.

Music played was:

    B├śRNS – 10,000 Emerald Pools

    Sandy Denny & Thea Gilmore – Sailor

    Jay Dee – Strange Funky Games And Things

    808 State – Spanish Ice

EL84 amp build thread here

Tony then used Mike's amp and speakers with his RTZ DSD source which comprised his LnP Signature CDPro2 CD player he'd made some years ago, the S/PDIF was passed to his much modified Transporter which he was using to take I2S to the RTZ DAC

Music played was from The Alien Jams by Ajja & Cosmosis

And a very ropey shot of the inside

Next was Steve (no. 1) with PX25 SET and speakers comprising Beyma TP-150 AMTs, Fane 12"s and Fane 18"s

The next double act was Nick and Dave. Nick had brought the latest incarnation of the turntable he's been working on for some years, now with super heavy delrin-skinned platter. Amp was a Luxman, and Dave didn't want to use the speakers he'd brought so they used Steve's.

An Ortofon arm and Denon DL-103R cartridge

1 o'clock brought the lunch interval, after which play resumed with Ray. Ray had travelled up some distance that morning but had brought his small office system with him. A 25W chip amp from abraxalito on DIYAudio and a Noir headphone amp for a touch more gain - the speakers, Ariston, £5 from a carboot, needed a little extra help in this room. This is the only decent photo I got. (The middle box was a passive pre with Muses volume, not used as maximum gain was required.

Music played was;

    Peter Gabriel - 'Rhythm of the Heat' from the album 'New Blood'

    Christine and the Queens - 'Tilted' from the album 'Chaleur Humaine'

    Be-Bop Deluxe - 'Adventures in a Yorkshire Landscape' from the album 'Axe Victim'

Chris had another go, this time with his small Fostex speakers and large Crown amp

Then Mike again with different speakers, BBC LS2s his uncle had given him

And that was the end of my day and I left the rest of the team to play on till stumps. Or beyond. A great day and everything sounded really pretty good. Sure there were flavours but everything worked. Thanks to everyone who made the effort to bring stuff.

Friday 26 August 2022

Design Of A Single Ended 2A3 Valve Amplifier - Part 2

In part 1 we looked at design of the 2A3 output stage of this amplifier, that was a very, very long time ago. Now we'll look at the D3a driver.

The Driver Stage

Output Stage Recap

Design of this driver stage is very similar to the output stage, but with slightly different goals. The purpose of the driver is to drive the output valve's grid, so let's have another look at the 2A3 anode characteristic curves to remind ourselves - here's the datasheet graph to which I added the safe operating zone.

If you remember, our operating point is 250V and 60mA, with a grid bias of -43.5V.

As this is a class A1 design the 2A3's grid should remain negative at all times (i.e. to the right of the red curve) so for full output the grid swings down from its bias point of -43.5V to 0V (and up to -87V). The driver is a common cathode stage, so the driver needs to swing + and - 43.5V on its anode, which is coupled to the 2A3 grid via a capacitor. Capacitors block DC voltage - we'll talk about this capacitor later.

Driver Stage Considerations

So how do we choose the driver? There are a number of considerations but if we want to use a single stage before the 2A3 then ideally we need to swing the + or - 43.5V from a standard 2V RMS input, such as a CD player. 2V RMS equates to 2 x square root of 2 x 2 = 5.66V peak to peak so + and - 2.83V. So in rough terms the driver needs a gain of 43.5 / 2.83 = 15.3. In reality a valve's datasheet amplification factor won't be fully achieved, maybe only in the region of 70%.

And we perhaps might prefer a little more sensitivity than + or - 2.83V, and + or - 1V gives us plenty of "welly" on the volume control. So let's aim for an overall gain in the region of 40.

It's not just a question of choosing a valve with the appropriate amplification factor, it has to be able to actually swing those volts too! And ideally we'd like it to have plenty of headroom i.e. some more swing in hand so it's not up against its limits and clip.

We also want to minimise distortion so might want to aim for areas of the anode curves with equally spaced grid voltage curves. (However, as the amplifier is the result of two stages, both stages have to be considered together as a combined system if we want to fine tune minimum distortion. But that would need some test equipment and is beyond the scope of this blog.)

And of course we need to obey the safe operating characteristics of the valve.

The D3a

So back to the D3a. It's a "special quality" pentode that was used for telecommunications. Here's the datasheet. Pentodes are generally newer than triodes (which is somewhat relative these days!) but can make good drivers as they typically have higher gain than triodes. Sometimes too high. But they can be wired as triodes with lower gain, and some think they can sound better this way.

Not all pentode datasheets contain triode connected info but fortunately the D3a does. Looking at page 3 the triode amplification factor is given as 77, so ~70% is a real world gain of about 50, which is nice. Now let's look at the second graph on page 7 for the average anode characteristics.

The curves look rather like the 2A3, but the values of voltage and current are very different. That's because this is a signal valve rather than an output valve.

The Operating Point

So let's look at the operating point and start with the safe operating requirements, see page 5 of the datasheet. Helpfully the maximum anode power dissipation of 4.5W is already shown on the datasheet graph. In addition we want to stay to the right of the 0V grid line, and not exceed the maximum anode voltage of 220V and maximum anode current of 30mA. So here's the graph again, with the safe operating zone shown.

Now we can choose our operating point. We want to swing 2 x 43.5 = 87V peak to peak on the anode, and we want a nice place on the grid voltage curves with equal spacing to minimise harmonic distortion. But also, unless we're going to have a separate power supply we need to bear in mind the B+ supply voltage that we determined in our output stage design.

Now if we go back to our 2V RMS input voltage from a CD player, which is 5.66V peak to peak, if we look at the curves above we can't adequately bias our valve for this swing - the grid voltage curves only go as far as -3V. In reality we would need a meatier driver.

But it's not necessary to swing the full 5.66V - music rarely reaches the full 2V RMS. If we go for 2V peak to peak then the amp will have decent sensitivity. So let's bias the valve to the right of the 1V curve. This does mean of course that there is the potential danger that at full bore the amp could overdrive, but in reality the volume control would never be turned up that high to permit the full 2V RMS swing in a well balanced setup system.

Now it's an iterative process of looking at operating point and anode loadline. An operating point of 150V and 20mA, which is a grid voltage of approximately -1.38V, looks a decent compromise. You can see I've sketched in the approximate -1.38V line at the operating point.

Loadline and Anode Load

When we looked at the output stage we went with the datasheet loadline of 2500 ohms, which is approximately 3x the anode resistance. Our choices for a driver are slightly different as the higher the load resistance generally the better as it will tend to reduce distortion. But the higher the load resistance (i.e. the larger value of the anode resistor) the more voltage will be dropped across it, which in turn means the higher the required B+ voltage.

The datasheet for a triode connected D3a gives the internal resistance as 1900 ohm. If we apply the 3 x rule: 3 x 1900 = 5700 ohms. So we're looking for a resistor of at least 5700 ohms, and ideally we want to use a standard value with suitable rating. So 6800 ohms looks pretty good, and the power dissipated in it will be the current squared x the resistance, so 0.02^2 x 6800 = 2.7W. We derate by 3 x to 5 x, so we actually want a 6800 ohms resistor rated between 8W and 13W, maybe a nice 12W Mills again.

The loadline is shown in magenta, and the thicker section shows the swing about the operating point - up to a maximum of -2.56V where the maximum allowable anode voltage of 220V occurs. As the grid voltage can swing up from -1.38V to -2.56V (1.18V) it therefore swings down to -1.38 - -1.18 = -0.2V.

Now if we look at the anode voltages at these points on the grid voltage we can see the anode voltage will swing up to 220V and down to 75V i.e. -75V and +70V from the operating point. Ideally those values would be the same as the difference reflects a little distortion - that's the consequence of unequally spaced grid lines. But a difference of 5V is pretty good in reality.

If we now think back to our 2A3 we need + and - 43.5V swing on its grid, and the D3a as we have it configured here can swing +70V and -75V, more than actually required. But as most music is recorded with less than 2V RMS, in reality we'll be able to use more of the volume control, without needing a pre stage. And if the music is too loud so the driver stage starts to clip it will sound a little unpleasant and the volume will inevitably be turned down.

Cathode Resistor

Nice and easy, we want to raise the cathode to 1.38V to achieve the correct bias and there will be 20mA through it, so using our friend, ohm's law, R = V / I = 1.38 / 0.02 = 69 ohms. Let's say 68 ohms as a standard value. Power dissipated through the cathode resistor = 0.02^2 x 68 = 0.03W, so we can happily use a half watt rated resistor.

Cathode Resistor Bypass Capacitor

Right, time for some more maths again. First let's work out the cathode resistance (check back to the 2A3 page for more explanation):

rk = (ra + Rl) / (mu +1) = (1900 + 6800) / (77 + 1) = 112 ohms

So R = 1 / ((1 / rk) + (1 / Rk)) = 1 / (1/112 + 1/68) = 1 / (0.0089 + 0.0147) = 42.4 ohms

As this is a driver stage and earlier in the reproduction chain we should aim for a much lower f-3 frequency than the output stage as once the bass frequencies have been lost they can't be magiced back. 1Hz isn't unusual.

And therefore C = 1 / (2 x pi x f-3 x R) = 1 / (2 x 3.142 x 1 x 42.4) = 3754uF.

Which is a big capacitor. But I had some Panasonic 2200uF electrolytics which gives an f-3 of less than 2Hz. Still lots of uF, but not unreasonable.

As the voltage at the cathode is approximately 1.38V a capacitor rating of 5V is fine.

Grid Leak Resistor

The datasheet suggests 500k ohms as a maximum, I went with 100k ohms, 1W is more than adequate.

Grid Stopper and Screen Tie Resistors

There are two more resistors that are very important. The first is called a grid stopper and is soldered as close to the valve socket pin for the control grid (G1) as possible. This stops valves oscillating. Higher gain valves, like the D3a have a habit of oscillating which is a bad thing. You might not audibly recognise oscillation but it will affect the sound to some degree. Sometimes it can be present but at very high frequencies, much higher than we can hear, but it can still have an affect on the music. A value somewhere between 300R and 2200R usually does the job. I used 1000R, and half a watt will be adequate.

The other resistor ties the screen grid (the middle of the three in a pentode) to the anode. This is what make the valve work like a triode rather than a pentode. 1000R half watt again.

Capacitor Coupling the Two Stages

Now we have the basic design of the driver stage we need to consider how we couple the driver stage to the output stage. The signal is taken from the anode of the D3a so we know the voltage there is 150V (our D3a operating point) plus the voltage we have raised the cathode by (1.38V) so ~151V. If we were to attach the anode of the D3a directly to the grid of the 2A3 there would be 151V on the 2A3 grid, which would be a very bad thing. Remember, from our 2A3 design there's 0V on the grid, -45V relative to the cathode, so applying 151V wouldn't do the 2A3 any good at all.

So we need a way to ensure this doesn't happen. The three usual methods are capacitor coupling, interstage transformer coupling, and direct coupling. Capacitor coupling is the most common, and for good reason; it's the easiest to implement. And whilst some say other methods sound better others are inclined not to agree; you need to build for yourself and see what you think. Capacitor coupling works in this application because capacitors block constant voltage but allow changes in voltage to pass, so using a capacitor blocks the 151V but allows the signal to pass to the 2A3 grid. We just need to work out how big a capacitor we need.

Whilst almost any capacitor would block the 151V on the D3a anode the coupling capacitor forms a high-pass filter with the 2A3 grid resistor, so if the value is too low the f-3 corner frequency could be too high and we would lose our bass again.

The high pass filter capacitor value is given by:

capacitance (farads) = 1 / (2 x pi x corner frequency f-3 (hertz) x resistance (ohms))

C = 1 / (2 x pi x f-3 x R)

Using the 40Hz f-3 we designed the 2A3 stage for gives:
C = 1 / (2 x 3.142 x 40 x 100000) = 0.4uF

0.4uF isn't a common value, but 0.47uF is. And it has the advantage that as it is slightly larger it drops the f-3 frequency to around 34Hz. Whilst beneficial as it drops the high pass filter a few Hz it also doesn't compound high pass filters at the same frequency, which wouldn't be ideal in some respects.

But you could also choose to use a 1uF coupling capacitor which would drop f-3 to 17Hz. This is perhaps a nicer place to be, but it requires a bigger value capacitor, and some audibly prefer smaller. More capacitance generally costs more too.

There is roughly 151V across this capacitor so the capacitor needs a higher voltage rating. Generally, you want to be using a nice film capacitor here. Some swear by paper in oil, others prefer PP. But most film capacitors used in valve amplifiers are rated for 630V, some 250V, so more than adequate.

Volume Control

This just leaves the volume pot. There is a balance to be struck here, 50k is pretty good but others can be used.

And now it's confession time. It's so long since I first started preparing these two blogs I've lost the drawing file somewhere on an old computer which had the circuits and values :-(. But if you've made it this far then you have enough persistence to fill the values in for yourself :-). So no final glory shot of the full amp circuit, but who knows, maybe in another seven years I might do a blog about the power supply...

Sunday 2 June 2019

Audio Meet - Owston 1st June 2019

Another year has passed since the last Owston meet, and my last blog. I'm not sure a post a year is really a blog...

Anyway, it was great to meet up again with lots of old friends and some new ones. The weather was fortunately dry (on Saturday at least) and warm if a little humid. Outside was a pleasant relief from the room inside which later on started to feel like a sauna with so many bodies, and so many kW being drawn out of the wall.

After lots of setting up and chat our "host" and organiser Steve welcomed all and got us underway.

This year there were a couple of dedications to Richard Dunn of NVA who had recently passed away. First Dave MCRU...

...then Lindsey who had attended Richard's funeral two days before.

I only met Richard once, the previous year when he attended Owston for the first time and I picked him up and dropped him off at the train station. He was an interesting man, and very different in person than his internet persona.

First away today was Ed with a system he explained was where he was 10 years ago. The 811 amp he had dragged out of the loft especially for the day, which he used with his own design VoFo speakers. The amp outputs less than 3W and as the VoFos aren't hugely efficient the audience listened in well observed silence. Source was a netbook with small USB DAC I think.

Steve C was next with his well regarded 13E1 OTL and Fane single driver speakers. The OTL makes less than 1W, but the Fanes are efficient enough to enable a decent volume in the room. I've heard this system in Steve's house a few times and it's truly wonderful, though I found the bass a bit light for my taste. But in this large room the bass was a very pleasant surprise, deeper and more extended. The combination makes really nice music.

Did I say I like valves that glow? They're beauties :-)

Then Ant, Steve's son, demoed his turntable with Steve's amp and speakers. This is one of the earliest of Ant's Lenco customisations, possibly his first, where he cuts down the top plate to resemble something along the lines of a PTP.

This time Ant also had a new unipivot arm he'd recently built from scratch, turning the pieces on his lathe. Sounded really very nice. Unfortunately I don't have any closeups of it - the room is tricky for photography, and much darker than it appears in the photos. The camera autofocus often struggles, and manual focus is tricky as it's so dark. Oh well.

And then it was me. When I've brought stuff before it's sounded somewhere between okay and horrible, so it was with much trepidation I tried again. It might have been a little unwise but the whole system I brought was being used together for the first time. What could possibly go wrong???

Well, it worked, as in it made a sound, and I thought it was okay, but was a bit too concerned with trying to get the bass balance right to really listen. I was pretty relieved as when I'd brought the Quasar speakers previously they had sounded dreadful as the bass wasn't attenuated enough at the much higher levels in the large room than at home. An important lesson was learnt that day.

But there weren't really many comments about how it sounded, apart from Dave's "can't polish a turd" (which was hopefully tongue in cheek!) so I guess there's more work to be done.

There are a few blogs that need writing about this kit, but the turntable is one of Ant's Lenco GL75 modifications with a Linn Ekos and Ortofon Kontrapunkt B. Neither would have been my first choice companions but I had them lying in a cupboard so it was a zero further cost option. The phonostage is a Salas' simplistic folded NJFET.

Then it gets a bit more interesting. The GM70 amp using coppers is directly coupled to the wideband Fostex 208ESigmas, with a simple first order filter to the Fostex T90A supertweeters. Although I roll the bass off via the GM70 cathode bypass caps so it's not stressing the OPTs I'm largely using the natural roll off of the 208 on the baffle at around 150Hz.

Below the 208s are Supravox 285GMFs in H baffles which are directly coupled to two cheap ebay 3116 class D subwoofer amps with a built-in lowpass filter at 180Hz. Not ideal but dead easy and dead cheap.

And a significant increase in clarity when I initially removed the Quasar series crossover to try this arrangement. These baffles were knocked up recently so are a work in progress, but I prefer them to the original Quasar.

Then came Stephen aka izzy wizzy, with a rather large van full of gear. An excellent effort for his first time visit :-). Hopefully it won't be his last.

It's tricky to know where to start with this setup. Speakers are Tannoy 15" Gold in the original 1950s corner cabinets. They're quite big! And rather magnificent. The turntable is a DIY job (from a group buy???) with Stephen's long term evolution phonostage. There may be a pre amp in there, or was it a headphone amp, or both, and then his 813 PP Nemesis amp. PP isn't my thing but I think this is the best one I've heard. Tremendous grip but with the subtlety and detail that thoriated filaments can give.

Stephen had specially made some sides to the amps as a safety feature. They rather reminded me of the art deco architecture of Napier in Steven's home country.

The blue glow is of course from a mercury vapour rectifier. All the amps used them :-).

Just a small box of valves...

And safety first of course...

Next was Chris. Following the big sound of the Tannoys wouldn't be easy, and the small Fostex speakers really didn't look up to it. But of course Chris had a trick or two up his sleeve, and paired with a small sub, a large solid state PA amp and some digital tricks I don't understand filled the room.

Chris has "adapted" the Fostex with a crossover. Astonishing performance really.

After lunch it was Colin's turn at the top of the room with his latest version of Eddingdale speakers with high quality drivers. Powered by Nick's original P6 solid state monoblocks and Njoe Tjoeb CD player. Personally I prefer baffles and efficient paper drivers but these are probably the best box loudspeakers I've heard.

And then it was Steve's turn. Steve was having a bit of a bad day as his PX25 amp had decided to have a strop and not work properly. Steve thought it might have been a dead rectifier, but fortunately had brought his little NVA amp as back up, so used that instead.

Next was Mike, who drives up on his motorbike and sidecar so travels light. Source is his phone, via Bluetooth I think, into his PCL86 spud amp using SMPS PS. He had brought small ex-music centre speakers but pressed Colin's Monacor coaxial driver speakers in to service. Sounded really nice.

Then we had a triumvirate of Dave, Nick and Gary, with Steve's speakers.

The turntable is a prototype that Nick has been working on with a company from York. They've fabricated the turntable, Nick has been working on the software control side.

Nick explained how he'd started with speed control, but ended up with phase control too, and remarked how he'd ended up at pretty much SP10 territory.

The phonostage is one of Nick's, and Gary's PS Audio preamp was used as some gain was required to drive Nick's 211 amp.

And then Thomas, who'd spent a fair bit of the morning assembling and then fettling the latest "Tower of Power" incarnation, now sporting GK71s. Unfortunately the DC heater supply had blown so some running repairs were required.

Thomas also brought his new Quasar baffles made from old wardrobes with Lowther PM6C and Eminence Beta 15 bass helpers.

Thomas seemed a little unhappy with how the system sounded, particularly the bass but I thought it was alright. I was a little surprised how well behaved the treble was, no Sidcup screech that I could hear.

There had, however, been a lot of unnecessary and frankly infantile comments about his Lowthers. I'm really not sure why...

So, next we were treated to JohnG's Quad ESL 57s, with the same turntable/ phonostage/ pre/ poweramp the Dave, Nick and Gary used. Apparently these had been refurbished reasonably recently, and though I haven't heard many electrostatics those that are familiar with them thought these to be very good.

If I lived in a world of Ella and Frank all the time these might just be perfect. But I don't and they don't have enough bass for me. But if I could have a pair in a room for just those moments that would be very nice indeed :-).

Then Ed demonstrated his second system with lateral FET amp and Ariel speakers (the piano black ones right back at the beginning). I've heard Aerials sound a little underwhelming but was impressed with these that Ed built.

I was starting to get a little jaded by now and spent some time outside. But I very briefly caught Les of Avondale Audio's amp with Colin's Eddingdale speakers, but didn't hear enough to be able to really form an opinion.

Then we heard Gary's PS Audio pre and DAC with Ali's 300B OTLs through Steve's Fane speakers. Very nice I thought.

And then we were treated to Mark's one treble horn. This was hardcore stuff for the purist and involved one channel of castanets amongst other things.

By this time I was struggling from all the different sounds we'd heard and it was time to pack all my stuff back in to the car. I did get chance to hear Ali's NCore 400 monoblocks and they were really very good. Food for thought.

And that concluded another great day. Everything sounded really very good, and there was lots of good chat. Roll on next year.