Help on assembling an Aikido 5687 kit

[Nick]

I've now written more posts in this one thread than in my entire history of reading HiFi fora, and I would feel a bit less like an exhibitionist if other people were to chip in if they're interested in the topic.

You are doing just fine

[machtman]

If you've never seen a valve data sheet then it's a good place to start, link here courtesy of everyone's favourite, Franks:
(deleted for users < 5 days)

Don't worry too much about understanding everything at first. At the end are some characteristic curves that are really helpful when designing, but you don't really need to understand them at this stage - the Aikido is very tolerant. Further up on the second page are three typical operating conditions (120V, 180V and 250V on the anode (or plate as it's an American sheet)). Below these are typical operating points - the manner in which the valve should be set up to work if you like. You don't have to stick with these, you can go off piste any time you like, they're just handy reference points.

With the Aikido each channel has four valves - usually dual triodes are used, like the 5687, so there are two valves per channel. The first valve is the gain/amplification stage, the second is the white cathode follower which is a load of jargon which means that it's a buffer (and is pretty much unity gain) but does a good job of driving the next stage. This is the bit you're hoping will improve what you already have. And it might, or might not, Nick's stuff is very good.

Back to the first stage, the gain from an Aikido is roughly half of the triode's gain because the two are stacked. Hence the 5687 Aikido has a gain of 8ish.

As Broskie says at various points when he's talking about the Aikido the circuit is exceptionally tolerant of the values of components used. But there are some better options than others. Fortunately JRB produces some fairly good instruction manuals. They are, however, not for complete beginners as even though he describes some stuff in great detail a certain amount of knowledge is required.

So if the above makes a bit of sense the next stage is to have a look at his PDF:
(deleted for users < 3 days)

On page 7 he gives 4 typical setups, you choose any of the four with confidence of good results. Which would you prefer? You'd only find out by building all 4. And having a memory good enough to remember what each sounded like.

I suppose I should check to see if any of this makes any sense before going further, but the next question is did you just get the PCB or did you get kits with most of the components supplied?

Hi Simon,
I'm back, only slightly less confused.
Point 1 no confusion: My Hammond 266N12 heater TX is 12.6 V so that is the option I'm stuck with.
Point 2. minor confusion. My Hammond 369AX power supply has outputs that look like I can readily produce 250 V (red to red) or 175 V (violet to red) or 125 V (red/yel to red).

The pdf you linked to is identical to the manual that was shipped. The 4 configurations on p. 7 aim at a B+ voltage (presumably at each 5687 or total???) of 100, 150, 200 and 250 V. None of those correspond to the load diagrams except for 250. On top of that all the load diagrams are for 6.3 V heater voltage. That may not make a difference because p. 7 specifies differences in Heater current according to heater voltage but everything else is the same. So if I were to use the load diagrams and try to calculate everything it would have been simplest to go for 250 V.

What finally brought a glimmer of hope to my overstrained and limited intelligence was that the Broskie blog that I referred to above from 09 Jan 2012 included his experience in building a 5687 all in one board. Firstly he specifies that he bought some Pearl Audio cryo-treated Tung Sol 5687WA valves together with Pearl Audio coolers. He found these were better than Philips JAN 5687's from the 1980s, which is the standard valve at Watford Valves. Pearl Audio coolers are available in the UK but I didn't find the Pearl Audio cryo'd Tung Sol's. But they are available at Parts Connexion in North America, and they are having a 40% price reduction at the moment because of over-stocking. Price per valve before reduction and shipping $94 for ultra-low noise and quite a bit less for low-noise. So I've sent off an order for a matched quad of 4 plus coolers and am waiting for confirmation of the price. If this Aikaido ever gets built, I'll be happy to report on those valves here. In the past, an NOS NIB TungSol 5687 from the US sounded quite a bit better in Nick's phono stage than a JAN Philips.

Sorry but I thought the above might actually interest more people than what now follows. In the rest of the blog, he also describes that he put it together with unspecified industrial TXs that provided 12 V for the heater and 150 V DC at the valve from 120 V AC and 170 V DC secondary. This doesn't match the recommendations on page 7, where 150 V B+ corresponds to 230-250 V AC secondary. It sounds like he used a different combination of resistors and capacitors than in the current manual which is from months later. So I assume I should trust the current manual.

He also makes some enlightening comments about goals for a 5687 Aikaido, including 'we must strive for the lowest raw B-plus voltage possible'.

Given that I can't provide 150-170 V Ac at the TX (and assuming that 175 is out of range) the lowest value I can provide is his second choice: 150V B+, 230-250V AC secondary giving 60 mA current, heater 12.6 V, Heater current 1.25 A.
This involves
R3,4 390 R, R6,7 240 R both of which he supplied

So assuming that there aren't any really convincing arguments against this choice, I would go ahead with it.

Now comes the next source of confusion. Broskie gives the options of wiiring the heater power suplply board for either a full wave bridge rectifier circuit, a full-wave voltage doubler rectifier, or as a full wave center-tap circuit. This is outside of my understanding. I don't know what these are and I don't know the advantages of any of them. Can somebody who has gotten to this point possibly share their wisdon?

Mark

[Nick]

Rember:

1. You have a seperate heater transformer, ignore the 6.3v secondary on the B+ TX

2. Ignore the bias tap on the TX, you have either 250v full wave, or 125v center tap.

[simon]

First thing Mark, DON'T PANIC! It will all become clear enough. You're asking a lot of questions, understandably, but take it bit by bit.

Don't get too hung up on the actual operating point of the valve (its setup). The only way you'll find out which you prefer is to try a number. But the Aikido is pretty tolerant so get it working first. For what it's worth I started with Broskie's 150V 7mA, but settled on 300V 15mA. But that's in my application, my ears etc. etc.

The usual way of designing a valve stage would be to choose an operating point (voltage across the valve, current through it and the bias voltage on the grid) based on some engineering choice, and then work out the required HT. From this a mains transformer can be selected.

But just as valid is to start with a mains transformer, as you have done by buying the 369AX, and working back.

The 369AX has what's called a centre tap. This means that the two 125V secondaries can effectively be joined together to form a single secondary of 250V. Using a centre tap is the traditional way as this would use a valve rectifier as only two diodes are required. These days we have solid state diodes and they're cheap as chips so four can be used to create a bridge and the two secondaries can be joined together to form the one 250V winding.

So you have the choice of either 125 x 1.4 = 175V or 250 x 1.4 = 350V raw HT.

A quick word on heaters. You can run 5687s at either 6.3V or 12.6V depending on how you wire up the socket. With JRB's all in one you don't have the choice though because of how he's configured the PCB - it's 12.6V. Or 12V depending on which resistors you use. I really wouldn't sweat this choice, anywhere between 12V and 12.6V will be fine. You may well not hear any difference whatsoever, and any difference you might hear would probably be swamped by other factors.

So the Aikido circuit. Basically the first stage is two triodes, one stacked on top of the other. So when Broskie talks about 250V for instance, that's across both valves, so there would be roughly 125V across each triode. So if you wanted to look at the curves you'd be looking somewhere below 125V (you need to take off the voltage dropped across the cathode resistor). Conveniently the manual provides values for 250V.

But you don't really want to drop ~100V in the power supply, it's not best use of resources, not that it matters really. So you could try to work out values for an HT of 300V, or a little more.

Or you could try the 150V values he gives. Which would mean dropping only ~25V in the power supply. I think actually I started with this operating point IIRC, but preferred more voltage and current. You might not.

This operating point draws 17mA per channel (7mA in the first stage and 10mA through the second). So to drop 25V across R17 and R12 you simply consider 34mA across R17 and 17mA across R12 (as each channel has its own R12).

Voltage drop across R17 and R12 = 0.034xR17 + 0.017xR12 = 25V

Plug in some combinations of resistors and away you go. There are more considerations, like power rating of the resistors, and even frequency cut off, but one step at a time. Let's determine how you want to set it up and we'll work through the important stuff.

[machtman]

Thanks Nick. BTW I got in touch with Ant and asked whether he would be willing to build a wooden box for the variac and also build a wooden case for the Aikido.

Thanks Simon. It's 04:20 and I'm off to the airport in an hour to spend a day at Warwick so won't be back in touch till this evening.
One request for clarification. Does the center tap you are not recommending correspond to the full wave centre tap that Broskie describs? And does the four diodes you describe correspond to the double rectiier he describes?

I need to study the load diagras again this evening and reread your post about 5 times before I can say more,

This is helping even though I am so thick

Mark

[simon]

Not thick Mark, just doing something new that some people find a bit intangible. Bruce Rozenblitz explains on the first page of his book how a lot of people struggle with electronics as they can't physically see the stuff - it was a lightbulb moment for me.

There are a couple of pages here that might help with the jargon of power supplies:
http://www.audio-talk.co.uk/archive/5c007.pdf
http://www.sowter.co.uk/rectifier-trans ... lation.htm

The beauty of valves is you can make it as complicated or as simple as you like. So you can go with the centre tap for 175V (which requires two diodes as your centre tap is effetively your earth connection), or you can join the two secondaries together for 350V (and use four diodes as you have to create your earth because there isn't one on the transformer).

There's no special magic sonic benefit to either choice, they're just engineering decisions. For instance, assuming the same HT, lets say 175V, a centre tapped transformer only needs two diodes but has two 125V windings.

On the other hand a single secondary requires four diodes but only has one 125V winding.

These days simple solid state diodes cost pence (and don't require heaters), two transformer windings on the other hand cost a lot more.

Before solid state a bridge would have required two valve rectifiers to form a bridge so the value equation was a little different.

You have a transformer that allows you to choose either 175V or 350V which is great . It doesn't really matter at this stage which you choose, you don't know what either sounds like so how do you know which is best for you? Yeah, if it was me I'd go for 350V HT, but I've tried both and there's no way of knowing which you prefer without trying both.

Hope you don't think I'm being oblique, just trying to help you understand that it's as broad as it's long, and the important thing is that you make the decisions. Otherwise it would be my kit.

[machtman]

Hi Simon,
whereas my first impression was why do I need to worry about all this, and can't simply find a kit that says solder this and solder that, I'm starting to enjoy this as a learning experience. Thanks for the links. Need to study them.

I discussed this with Nick. You say 300 V, Broskie says 75. Nick would prefer 240V. That's what happens when you get 3 opinions. But he also came up with a solution which makes great sense to me.

Full quote:
"
me>: What do you think about the Aikaido? Should I go with Brioskie's reasoning and use 150 V or with Simon's experience and go for 300?

nick>: Do both, and then use what you prefer. You can wire your variac between the incomming mains and the B+ trasformer. Wire it for 300v with 240v mains, then you can wind it down till its does what you like. For me, 75v across a 5687 is a little low, personally I would have gone for 240v instead of either, but of the two, 300v is closer to what I would pick myself.
"

Trusting my own ears rather than theory makes more sense to me. So I think I will connect the variac, which is now in my hands and weighs 2.5 kg and still needs a case and connecting it to the B+ transformer sounds like a wonderful idea.
Nick also recommended buy some one sided PCB pins, soldering the desired resistors and caps to them for each voltage to be listened to, and then choosing the one I like best afterwards makes enormous sense as well. I just need to leave the leads to and from the B+ long enough that they can be soldered again after everything is mounted, while avoiding electrocuting myself.

So now we get concrete. I can test all the combinations calculated by Broskie if we wanted or simply fix on 150 V from his list and calculate new values for which I have to buy resistors and caps for 75 V, 240 V and 300 V. I stress we for this decision making process because if I am going to do all that soldering and de-soldering, I would hope that you and Nick (and any lurkers who want to join and post something) would come to a central point (preferably my home) and share in the listening comparisons. And rather than we, I would much prefer if YOU could calculate the non-Broskie values because I am uncertain where you got the expencted amperage from. And the I can order one sided PCB pins, resistors, caps and WE can start reaching other decisions on which configuration to use.

This is actually starting to be fun, except that I am multitasking too much because 30 NM ECM Keith Jarrett and Jan Garbarek LPs have suddenly arrived from the US where I bought the lot from the son of a (deceased???) audiophile collector for $5 each plus $80 shipping. They look unplayed!!! Bit they still need washing on an RCM and my Loricraft costs me about 1 hr for each 3 records. And I need to get my second nixie fixed so the third pair of LEDs actually glow blue and the parts have just arrived. As well as measuring all the components that came with the Aikido board.

This is fun!

Mark

[simon]

Well now you're just spoiling the fun by jumping through things . I'm an engineer dammit and you have to do things in the right order!

Okay, let's cut to the chase then.

I was saving PCB pins for later, but this is exactly the thing to do if you want to try different components. Broskie's PCBs are the best (of the relative few) I've used and would stand resolder two or three times no problem I'm sure. Unless you're heavy handed. Using the pins means you're only soldering to the board once which is best, but I'm sure Nick's already told you this.

I prefer 300V, Broskie suggests 150V (not 75V, that's across each valve - remember it's an Aikido), Nick 240V. So that's 75V, 120V and 150V across each valve. I tried 100V too, but preferred 150V. I didn't try 120V. As Nick says, if you build for 150V and use your variac you see what the difference is like. Just make sure your heater transformer isn't also connected to the variac.

JRB has already given you 4 typical setups in his manual for 50V, 75V, 100V and 125V. So you could just use 125V and aim for an HT of 250V, but then you're dropping 100V in the power supply.

So to work out your approximate operating point for 150V you need to look at the curves. Morgan Jones is pretty much the definitive text on this so it's worth reading and re-reading this bit until it starts to make some sense. It's the foundation of valve design and understanding.

So let's say I'm looking at 150V @ 15mA, I'd look at the anode curves (which JRB helpfully puts in his manual). Actually I'm going to look at just under 150V as I know I will have some voltage dropped across the cathode resistor, let's say 145V. At 145V and 15mA the grid volts is just under -6V, say -5.8V. This is known as the valve's bias and controls how hard the valve works.

To drop 5.8V @ 15mA a resistor of 5.8 / 0.015 = 386 ohms is required, so we can use 390R.

You can see that with a voltage of 150V minus the bias voltage of 5.8V gives a voltage of 144.2V across the valve. I started with 145V, so close enough.

We should check the power rating of the cathode resistor: power = IV = 0.015 x 5.8 = 0.087W. You should multiply that by at least 3 for safety. I tend to use 5x (cos I'm anal) which would be 0.435W, so a half a watt rating is fine.

If that makes sense we can sort out the power supply.

BTW Nick, I'm interested in why you ended up at 240V - presumably you tried various and preferred that, or were there other factors too?

[Nick]

BTW Nick, I'm interested in why you ended up at 240V - presumably you tried various and preferred that, or were there other factors too?

It was a compromise from looking at the curves. I actually would pick 320v if it was left to me, I like the 5687at 15ma 160v, or even 180v. But given that JRB suggests 75, I was looking for a lower voltage, 120 was the lowest I would have tried looking at the curves, 75, seems far to low for me, unless its only a few ma.

[simon]

Fair enough, we're not that far apart then. I agree 75V looks way too low on the curves though.

[machtman]

Where did the magic 15 mA come from????? And I have no problems wit
-5.8V bias but where did that magic number come from. Apparently this is a fixed bias and unlike my amp where I adjust the bias with a pot and an LED here it is hard-wired and I just assume it never changes apparently.

Otherwise makes perfect sense. I'll see if I can work out the other resistors myself if all are supposed to run at 15 mA.

The second nixie is now working. All LEDs on and I had to desolder a few things and the desoldering gun is working fine whereas desoldering braid is horrible. It was useful experience because the kit supplier had insisted that I identify the problem by measuring the resistance across one resistor and ahorting the leads alternately on the two LEDs in series which were both not working. Trying to measure resistance through soldered connections was hit and miss, more miss than hit, but I convinced myself that there was the correct resistance in the end. And shorting two LED was the toughest soldering I had ever done. But in the end I got it done and found that one of the two LEDs was the problem. After replacing it the LEDs all worked. So I got ambitious and desoldered two other LEDS where the leads were too long, which also worked fine. Put everything bacl together and the valves didn' light at all. Staring at the PCB showed me that somehow I had inadvertently put it down somewhere, bending the vertical resistors and shorting two of them against each other, After straightening them eveything looked fine and it worked as it should. So I guess I am heavy handed and should use PCB pins, especially because there is resin all over the nixie PCB and it doesn't look as clean as I would like. But both nixies are working fine and I can post pictures if anybody is interested.
Mark

[Nick]

Cool. As for the magic 15ma. Its from the valve plate curves. There is a thread somewhere abouts that walks through the way those curves are used.

[simon]

Yeah, 15mA is just a choice, like 150V. From looking at the curves and taking a number of things in to consideration 15mA seems like a reasonable choice. And I found it to sound better than 10mA. There are many, many possible combinations, some are better than others.

[machtman]

Hi Simon
I've read and reread your posts and the manual and am starting to understand some of the terminology. I've stared at the curves and can reconstruct where you got voltage versus mA versus bias voltage for 145 V from. But when you talk about 390R across the bias resistor, which resistor are you talking about?
In an earlier post you calculated both R17 + R12 across each of the valves. JRB lists the R17 range as 100-1K but only supplied 19, 20, 100 and 200 1W. He gave the range of R12 as 100 to 20K but only supplied 1.6 K and upwards. So are you calculating the desired value for R17 and then R12 is calculated from your earlier formula for calculating both? Or is this the combination of both in series?

And is the 15 mA the idle current??? And if so, you used different currents for R12 and R17 in the mixed calculation, one of which was half the other. So is this 30 mA across R17 and 15 across R12? Still very confused about these details. But I have the feeling we are getting closer.

thanks
Mark
thanks
Mark

[simon]

The 390R is the cathode resistor - I don't have the schematic to hand but it's the resistors that sit under the cathodes of the 5687s.

Once you know how much current is being drawn by each channel you can size R12 and R17. The first resistor in the power supply (the one nearest to the rectifiers) is common to both channels so the current through it will be the whole current of the amp (2 x a single channel). JRB then splits to two channels (for some degree of decoupling between the channels) so this resistor has half the current through it than the last one i.e. just that channel.