September has come around again, which means that once again Andrew our Technical Director and my good self are in Plymouth for the Cell Physiology Workshop. Unfortunately this also means that my consequent escape from day-to-day Cairn activities leaves me enough time to start putting another blog together! This one is going to be a pretty mixed bag, but if there is to be a common theme, it’s going to be our Optopatch patch clamp amplifier. That’s more than a touch ironic, because only a few weeks ago we were thinking about not promoting this product any more, at which point another burst of orders came in!
So what’s the issue here? Well, the Optopatch has been around for a while now, and some of the components are becoming increasingly hard to obtain. It’s also very time consuming to build, which is a problem when everyone is already working flat out (maybe we should employ more production people – yes, just done that, meet Alan on our rogues’ gallery section!). When we were smaller, having a time-consuming product to build was quite handy, as it could always be built “for stock” if anyone was at a loose end, but that just doesn’t happen any more! However, “the market has spoken”, so a reprieve has duly been issued (sorry Dallas, just keep building!).
Actually there’s a rather close connection between the Optopatch and Plymouth, for two reasons. First, the Optopatch is based purely on analogue electronics, even though some of the things it does – notably the capacitance measurement system – you might have expected to be done via software on digitised signals. The link here is that a key part of the Plymouth course is a day spent building and testing analogue electronic circuits, albeit relatively simple ones. In spite of the rush to digitise everything in sight, analogue electronics remains as useful – and powerful – as it ever was, and those of us who know this consider it both important and useful to introduce it to others. Also, for anyone to genuinely understand what electrical recording, and especially voltage clamping and patch clamping, are all about, some sort of grounding (no pun intended!) in analogue electronics is pretty much essential anyway.
The second reason is that it just so happened that we first got an important part of the Optopatch, namely the optical current-passing system, working here during one of the earlier Plymouth workshops, way back in the mid 1990s, so we’ve always felt some sort of affiliation there.
For better or worse, the Optopatch has been one of my personal design projects within Cairn, so that begs the question of how and where I learned my analogue elctronics, as I was just a humble biology student myself once, so not unlike the ones who attend these Plymouth courses. This gives me an excuse to talk not just about my own student days, but also about my past interests in the subject of audio electronics, and hence of the part they played in the establishment of Cairn – which I have already indicated to be an interesting story.
To be honest, I’ve only recently realised that the “businessman” part of me was rather more deeply engrained than I’d previously appreciated. I’ve already related that my PhD project was a safe but boring one, so I’d ended up supplementing it by designing all sorts of electronic and similar bits and pieces to help it along and keep me (relatively) sane. What I hadn’t confessed to before was a parallel interest in audio electronics, especially audio amplifiers and loudspeakers, so I was spending some time messing around with those too. These activities developed to the point where I was getting some of my designs published in various electronics and audio magazines, and I was getting a fair number of other articles published too, which all gave a handy additional income to what would otherwise have been the standard impoverished student. I wasn’t getting rich, but more to the point I wasn’t getting into debt either! If only students could still manage that nowadays….
Once I took up that postdoc position in Boston, I had a proper research project to get into, and the Atlantic Ocean was rather getting in the way of those other interests too, so I just forgot them for a while. However, once I was back in the UK, I cautiously reactivated some of those old links, both out of curiosity and because I wasn’t sure about what my longterm future would be, to see if there was anything still worth “doing” on that front. As a result, I was introduced to a couple of people who had set up a business selling what we quaintly in the UK call “valve” amplifiers, although I prefer the more descriptive American term of “vacuum tube”, or just plain “tube”, so I’ll use that one here. I was told that the product was potentially interesting, but it had some problems. One story, although possibly apocryphal, involved an expensive carpet, a seriously overheating amplifier, and a rather hasty retreat! That rumoured story always reminds me of a wonderful edition of the “Only Fools and Horses” programme on British TV, which also involved a hasty retreat, in that case as a result of a misunderstanding over which of two chandeliers was about to be carefully released from the ceiling into an awaiting blanket (“Did we give you our address?” they asked, as they made a mad dash to their getaway car).
However, having been brought up firmly in the transistor era, there wasn’t much I could do to help them, and in any case by then I’d landed that apparently wonderful position at that old “research” lab that again I’d better not name. There I suppose things might have rested, except it turned out that the position wasn’t so wonderful after all, and the increasing realisation coincided with one of the people from that amplifier company getting back in contact with me, having set up another one on his own.
Although this was all a while ago now – we’re talking early to mid 1980s – I want to tread carefully, because the company that person had set up is still around, so some tact is still required! The following “observations” on the world of audio, especially its more esoteric extremes, are in any case rather general ones, rather than being aimed at any specific company, let alone that one. The fact is that I very much admired (and still do) the person in question, so I really wanted to help him, and in practice I ended up learning a huge amount about what it was like to run a small business. That turned out to be very useful for the establishment of Cairn of course, although I wasn’t thinking about it in that way at the time! Instead, I found myself getting more and more deeply involved in that business, and if circumstances had been different, we might even have been able to make a go of it together, but there were two difficulties. First, he certainly wanted me, but increasingly clearly as an employee than a partner. I was already an employee at that “research” lab, so was looking for something more! Second, I wanted to do something more “generally useful” to the world than making upmarket products that most people couldn’t afford (oops, I suppose that’s what Cairn does, but at least there’s the hope that our customers can do “generally useful” things with them). This was all going on around the time I realised that I needed to do something else with my life anyway (I can still smell that tosyl chloride, as related in a previous blog) , so in the end I had to tell him that I was going to set up my own business rather than to join his. Unfortunately, as can happen in such situations, our relationship then came to a rather abrupt end because that decision didn’t go down too well(!), but the last thing I wanted was to go into competition with him, so audio electronics has always been off-limits for Cairn – well, sort of, anyway!
And that brings us to the Optopatch. As the name suggests – although we’ve subsequently put the “Opto” prefix onto many of our other products – this amplifier makes use of an optical technique, in this case to generate the relatively small currents that patch clamps need to pass. The idea is simple enough, as you generate the currents directly by shining light onto a photodiode, albeit a very small one to ensure that this component doesn’t generate significant noise (which any “ordinary sized” one would certainly do).
Now, the relevant point here is that you need to be able to generate currents in either a positive or a negative direction, whereas a photodiode is fundamentally a unidirectional device. So, in order to meet that requirement, you need to have two photodiodes in opposite orientations, and to illuminate one or the other as appropriate. That’s fine if the requirement is to continuously pass current in one direction or the other, but what if the currents are small and repeatedly changing direction? In that case the directional changes have to be handled very cleanly, or the current waveform will be distorted.
Well, it turns out that audio amplifiers potentially have exactly the same problem, because in the standard so-called “class B” configuration (there are others, but they’re not so practical) they too need to use different devices to pass the positive and negative currents. It’s therefore important in this case too that the switchover from one to the other takes place cleanly, and if it doesn’t, the problem is known as “crossover distortion”. The trouble here is that any resulting nonlinearities will be most evident during the “quiet bits”, which is exactly where you don’t want them, because they’ll be relatively more audible there.
To be honest, this just shouldn’t be a problem nowadays with any half-decently designed amplifier, but in the earlier days that wasn’t always so, especially with some of the first transistor amplifiers. Perhaps that is why the view that tube amplifiers are “better” has become folklore in at least some circles, even though those designs have an even more fundamental problem, which is all to do with antimatter.
Semiconductors basically work by the materials having either “too many” or “too few” electrons in their crystal structure, which allows the surplus electrons, or the “holes” where there aren’t enough of them, to be available to pass a current (since they can both move around). To a reasonable approximation, a “hole” allows a current to be passed in a similar way to what could have been achieved if there were actually a few positrons hanging around instead, but without the inevitable risk of them meeting an electron, and then the whole device going off bang in an unusually spectacular manner. So, that allows the creation of “complimentary” semiconductor devices, which can be used as an opposing pair to pass current in either direction.
However, to get that to happen in tubes, you would have to use real positrons, which would bring in all the problems of antimatter containment, and in spite of the large sums that some people are prepared to shell out for their audio (more of that in a minute), I just can’t see that catching on. Nevertheless, the possibility of such an amplifier is an intriguing one. Amongst other issues though, it would always need to be sited with a direct view of the sky, so that the innards could be ejected into space to explode somewhat less dangerously there if the containment field broke down (just like they have to do in Star Trek if there’s a warp core breach). It would therefore be a definite no-no for most apartment dwellers for a start.
The saving grace for tubes is that compared with what the average loudspeaker needs, they tend to be produce rather higher voltages and rather lower currents than it requires, but that is fixed by using a coupling transformer that effectively swaps the extra volts for more current. Here, to pass a current the other way, you can simply reverse the transformer connections for that device, so having an appropriately wired coupling transformer driven by a pair of “ordinary” tubes means that you can actually pass current in either direction without needing positrons after all.
The only potential problem is that neither the tubes nor the transformers are particularly linear, but the nature of the nonlinearity (low-order harmonic distortion) is such that it tends to be spread out over the entire output characteristic, rather than being concentrated in that crossover region. Also, since it’s the harmonic components that at least partly give individual instruments their characteristic sound, such distortion can actually sound quite “musical” under at least some circumstances. The sight of the tube anodes pulsing red during the really loud bits is also something that no solid-state amplifier can emulate (they just go bang).
With transistors, the nonlinearity in practice tends to be more concentrated at the low-current end, so crossover distortion is relatively more of a problem. Like any other form of distortion, it can be reduced by negative feedback, which is a whole subject in itself. But briefly here, the amount of feedback that can be applied is related to the highest frequency that an amplifier can possibly pass, and to obtain a given level of feedback in the audio range, the “bandwidth” of the amplifier has to be proportionately higher than that, so suddenly we’re talking megahertz rather than kilohertz here. One of the problems with the earlier transistor designs was that their bandwidth tended to be rather lower than ideally required, and in any case excessive levels of feedback can introduce other problems of their own, so at least part of the trick is to try to ensure that the amplifier is as linear as possible in the first place.
Where I came into the story with that old friend was to be amongst the first to design an amplifier using a type of transistor known as a “power MOSFET”. As well as having linearity properties more similar to a tube than a “standard” power transistor, these devices also have higher bandwidth, allowing reasonable levels of feedback to be applied over a wider frequency range. Although further device improvements have subsequently alllowed others to achieve even better technical performance, my own opinion (for what it’s worth) is that at the very least there are diminishing returns here! That’s why I’m still using that same amplifier design for my own listening today.
However, we’re getting into a very contentious area here. I think it’s fair to say there is a continuum between legitimate and relevant technical issues concerning amplifier design, that merge seamlessly into the realm of the complete and utter fruitcake. The problem is where to draw the line, although I personally draw it as close to the legitimate technical end as I can. Part of the issue must be that if someone thinks that something will make their amplifier “sound” better, someone else will be very happy to come along and fill that need for them! That, I suspect, is why you can buy gold-plated mains power sockets (a snip at £129.95) or if you’re feeling really flush, you can get a rhodium-plated version for £165 – currently only two left in stock, so do hurry! Personally I prefer the rather “softer” sound of the gold ones, even though they are somewhat cheaper (sorry, I’m just trying to get into the spirit of things here). That will also help me save up for a nice pair of Nordost loudspeaker cables, mine for a mere £13,599.99! Oh dear!!! Well, I’m afraid there were signs of this nuttiness even back when I was involved. For reasons which were never entirely clear to me, back in those days there was one particular audio “reviewer” whose opinions were felt by my friend to be of some influence. I have to say I would have been rather more confident of his opinions if his “day job” had some connection with the physical sciences, rather than the distribution of home cleaning products sold door-to-door, but I felt I should keep an open mind nonetheless. After all, it was theoretically possible that he had an unusually acute sense of hearing, that would allow him to interpret any imperfections in what he heard, in terms of specific design or performance shortcomings. Perhaps he could say something like “Ah yes, there is clearly some thermal modulation of the output stage quiescent current”, or perhaps “The currents in the input long-tailed pair aren’t properly balanced, are they?”, or even “I think the open-loop linearity of the driver stage is a possible issue here”. After apparently careful listening, he would appear to prepare to say something along those lines, but the actual comments were always in the form of “Well, it’s really not quite there yet, is it?”. To me, that was about as much use as a chocolate fireguard, or in fact rather less so, because you can at least eat a chocolate fireguard, and I’m very partial to the stuff!
So, I’m afraid I don’t miss the world of audio. But while I’m on the subject, related to my increasing disillusion with it all was the then relatively recent introduction of the “compact disc”, or CD, which had not been as well received by many of the “audiophile” community as might have been expected. To be fair, and in retrospect, the quality of those early CDs wasn’t always anywhere near as high as it could have been, for at least two reasons. First, the methods used then to “digitise” what were originally analogue recordings weren’t as good as the ones used now. Second, the “master tapes” used for those digitisations were themselves often copies, or even copies of copies of the true ones, which introduced yet further deterioration. That has resulted in a nice gravy train for the recording companies, since they’ve subsequently been able to sting us all over again for “remastered” versions, for which the improvements can be very audible. Perhaps those old problems are one reason why so many people seem to think that vinyl is somehow “better” – even if (as is so often the case now) the recording was made digitally in the first place?
However, although vinyl has performance issues all of its own (personally I just couldn’t wait for the arrival of CDs!), I’m not going to be drawn into that argument, for a very simple reason, which is to do with trains. I don’t think anyone in their right mind could possibly argue that steam traction is fundamentally superior to electric, or even diesel, although the sight of a steam TGV, if only it could be done, would be impressive indeed! To me the truly amazing thing is that when nothing else was available, people got steam power to work so well. Now, they’re even building new ones, and quite right too. The real point for me is that steam traction, at some level, is just plain fun! Just like the glowing anodes of tube amplifiers, and just like vinyl discs, the world would be a far more boring place for their passing. That means that in some ways, they are all “better” than their later counterparts, so let’s not get too carried away here. It’s just that I much prefer CDs to vinyl, ok?
Ooops, I was carried away a bit myself there – sorry about that – so now back to the Optopatch. It turned out that here the crossover distortion problem has never been an issue in practice, for two reasons. First, there is lots of bandwidth available to apply sufficient feedback to prevent it from becoming a problem in the first place. Second, there is a slight “cheat” in that the higher-frequency components of the current are actually supplied via a small capacitor, which is bidirectional by its very nature. It’s just that you can’t pass a steady current indefinitely that way, unless (as with “integrating headstages”) you discharge it at appropriate intervals. In our case, the optical feedback can do this slowly on an ongoing basis, so it actually serves to prevent a significant charge buildup on the capacitor in the first place. So, all rather nice really.
However, for the Optopatch electronics, that’s just the tip of the iceberg! The real complexity – and fun – is in the resistance and capacitance compensation circuitry, specifically with respect to cell membrane capacitance measurement, for which we devised a very nice system, although it always seems to have been a rather well-kept secret!
There are various ways of measuring membrane capacitance, but basically they all boil down to the measuring the characteristics of a resistor-capacitor (RC) network, where R (you hope) is the resistance of the electrode that goes into the cell, and C (you hope even more) corresponds to the capacitance of the cell membrane. So you end up needing to measure both, even though you’re not so interested in what R is – in fact you want to ensure that the value of R, especially if it’s also changing, doesn’t interfere with the estimates of C.
A nice way to do this is to use a lock-in amplifier, in conjunction with a sinusoidally varying command voltage applied to an electrode inserted into the cell (or more usually nowadays, a patch electrode with a broken membrane seal). The amplitude and the phase of the resulting sinusoidal current in principle gives the necessary information to work out the R and C values. In practice, the amplifier’s cell resistance and capacitance compensation controls are usually set up so that all this current is actually passed by the “cell” compensation capacitor in the headstage, with this condition being easily found because the measured current trace is then completely flat. There is only one setting of the controls that can achieve this, and of course it gives the starting R and C values.
What one now wants to do is to measure changes in C, unaffected by R, and except in extreme cases these changes are going to be relatively very small. Any change in either R or C will cause a residual sinusoidal signal to appear on the current trace, because the R and C compensation controls of the amplifier are no longer “correctly” set, so how do we determine to what extent that residual signal is due to a change in R or a change in C? That’s where the lock-in amplifier comes in, because it can measure the residual signal in terms of two different components at 90 degrees to each other. If the “switching phase” of the lock-in amplifier is correctly set, this allows the residual signal to be measured in terms of its R and C components, because they are at 90 degrees to each other. The only problem is to determine what the correct switching phase actually is! If it isn’t correct, there will be crosstalk between the R and C components of the residual signal, which of course rather defeats the object.
There are various ways to do this, but the Optopatch does things rather differently, which gets round the problem in the first place. In the Optopatch, the residual signals (which are always measured as their “rectified” components by the lock-in amplifier to give DC signals) are actually used to provide control voltages that modify the amplifier’s R and C compensation settings, in such a way as to keep the residual current zero. So, by electronically measuring those settings over time, changes in both R and C can be followed exactly with perfect discrimination, because there is only one setting of the controls that will keep the zero residual current condition. The switching phase only needs to be approximately (rather than exactly) right to achieve this!
We were pretty excited about the success of this method, so we duly fired off a paper to Biophysical Journal, which is where a lot of membrane capacitance membrane stuff gets published. Unfortunately, it can’t have included enough maths (unlike most other methods, it doesn’t actually NEED any!), because they fired it straight back again without even sending it out for review. We were pretty sore about that, but of course what they publish is their decision. It actually got published essentially unchanged by the European Journal of Physiology so it can’t have been that bad really, although I think the Biophysical Journal would have had a more appropriate target readership, which is perhaps one reason why our method isn’t so well known. The people who DO know it certainly like it though!
Finally, when designing the Optopatch, the one thing I just couldn’t understand about patch clamping was the weird way that “current clamping”, which is patchclampspeak for the standard microelectrode amplifier “voltage follower” configuration, was being achieved by other patch clamps. This was resolved by a couple of people asking me if I was going to do anything about the “current clamp problem”, from which I deduced that the reason I didn’t understand the circuit was that it didn’t work properly! So, I came up with a configuration that did. The issue here was that in order to current clamp properly, you needed to reconfigure the headstage in a way that required some additional connections going back to the main amplifier box, but from the way that we’d implemented the optical current-passing system, we had those connections anyway! Therefore, although it was a relatively late design change, it was very straightforward to do.
There, that’s torn it! I’ve found myself talking about and actively promoting a product that we had just decided to demote to “legacy” status, so on top of the recent orders we now might even get some more! Perhaps we should put the price up to quell the demand a bit? In fact, how about charging £13,599.99? For that money you can have EITHER an Optopatch OR a loudspeaker cable. In fact, we can offer an even better deal. Give us £13,599.99 for an Optopatch, and we’ll INCLUDE a loudspeaker cable that will almost certainly work just as well as that other one. Everybody (apart from the people who make that crazy cable of course) wins!