Evolution: the gradual development of something into a more complex or better form
( Encarta )
I read Rudy van Stratum's (email: email@example.com ) article 'Transitors for the tubelover' in the dutch magazine 'Audio + Techniek' in the winter of 1998. It mentioned seven legendary amplifier designs which all used transistors to power up the sound. These designs came from the hands of worlds most talented audio equipment designers like Tim Da Paracivini, Nelson Pass and Hiraga. Also two designs of John Linsley Hood (JLH) were mentioned. This article got my hands itching to build something and I decided to build JLH's 1969 10 Watt design, because the components were easily (and cheap) to get, and its supposed quality.
Well this certainly didn't disappoint me, especially after upgrading some key components and fitting in a better supply (twice). Proud as a peacock I made some pictures of this amp, and published them on the Internet. Then I could never imagine the response I would get on this site. People from Tietjerkstradeel (NL) to Tokyo contacted me to tell me that they also enjoyed the JLH, to mention other websources or to ask for building tips. One day I received an email from Rudy van Stratum, the author of the article I got my inspiration from, to ask me what I thought of JLH's design. Rudy told me he was very curious about what would happen to the sound if the output elco was removed from the amplifier. His hypothesis was that the elco adds much coloration to the sound, so removing it would improve the sound quality. Sounded logically so we started to figure out how the schematic must be altered to make the removal of elco C4 possible.
First of all you need a symmetrical power supply, so the output signal can vary round the ground level. Rudy tried this but there was a big hum in the loudspeaker signal, probably inject via the suppply lines. So big resistors were placed in series with the supply lines, the hum was gone now, at the cost of some power (that was dissipated in the supply resistor). So this works! another problem was the big dc offset, and it took the amplifier a long time to get its output to about zero Volts. We found a possible error in the feedback circuit, the ground of C5 should be connected to ground in stead of V-. This would correct the dc offset problem. Rudy thought the sound of this design was even better than the original though! Then we couldn't find the solution of this problem and the project was forgotten.
Until Rudy read an inspiring story about the JLH in the spring of 2001 and he decided to give the old project another look. He contacted me that he had posted our problem of the JLH on the www.diyaudio.com site. On this site was a discusion group about the JLH 10W. The reply of Geoff Moss (visit his very nice site at http://sound-au.com/tcaas/index.htm) gave us some answers we already knew, but he also had the answer to the hum-issue! In the first stage of the original design, JLH added R1 to form a low pass filter in conjunction with C1. This filter keeps all the supply rubbish away from the first stage. So if you put it to work on two supplies, both positive and negative, you should also add this filter to the negative supply line! Eureka! This was the answer we were looking for. Also Geoff told us that the dc offset problems are known for this amp, the asymmetrical version also has this but it is not an issue there (due to the dc blocking of the output elco). Rudy build this amp with R14 as negative voltage low pass R, and it worked! Even the dc-offset is much smaller, and it stabilizes much faster. So this one resistor seems to solve all of our problems.
Things get better and better...
According to Rudy, the Evolution version of the JLH sounded better from the start. Much improvement through a very small adjustment! Geoff Moss did some simulation of the circuit and discovered a 'minor' error in the circuit at july 6 2001 he emailed me;
I was having another look at your JLH Evolution page and have noticed an'error on your schematic. The feedback capacitor (your C5) is shown with the incorrect polarity and will be reverse biased if connected as indicated. When the input transistor (your Q2) is biased via base resistors, its emitter sits at about -0.7V with respect to the output and earth. The emitter of this transistor is only positive when the dc offset is controlled by injecting current into the emitter as in the 1996 version.
Well, this minor error appeared to have quite some impact on the quality of the amplifier. I have never heard the amplifier with the incorrect polarized capacitor, but people who did assured me the change had quite some impact.
Another improvement can be the removal of R3, the small resistor in the positve supply line of the output stage. Rudy put it in to make it more easy to monitor the quiescent current in the output stage. I copied his schematic so I also had it in but bridged it on Rudys advise. Personally I can't really say I hear the difference between the amp with and without the resistor, but I haven't got the amp completely in shape (9-9-01). This resitor is quite easy for monitoring the current so I suggest you build it in the circuit but just bridge it when you don't need to measure the current anymore. Nobody has an explanation for the change in sound with and without R3. You would expect the quality to be better with R3, since it is an extra way to filter out noise from the supply. So if you have any ideas.. share the knowledge!
The most important aspect of an amplifier (for me..) is its sound. It came to speak in the text allready, the sound of the Evo version in comparison with the '69 design is much better. I didn't expect to be able to hear it right away but I did. The bass is much fuller, rounder and the complete sounding profits from that. Some people have said that it sounds as good (or better) as many tube amplifiers, but I have no experience with tubes. The '69 version allready was a good design, especially considering the ease to build, and the standard components used, but the Evo only makes it better.
Building the thing
Anybody who has this amp in the assym. supply version can very easily turn it into the symmetrical version. Just remove the output elco ,C4, make the filtering/buffering of the first stage symmetrical by adding C6 and R14. And of course change the supply for a symmetrical one, of 2x20Volt or lower (less dissipation). And finally reverse the elco C5! It might be smart to measure the output dc offset before connecting the speakers. You get yourself a better amp for the price of a new supply (and one elco/resitor).
Practice learns us that the amp has a few components which are very 'audible'. The one with the most influence is probarly the input cap. When I builded the amp I used a simple Wima MKS type. This sounded pretty good, but to tweak it a bit I soldered in an expensive cap of SCR, a PPE MKP type. (actually if you look at the pictures you can see the red Wima in the upper stage and the black SCR in the lower stage). This ruined the sound! The high became very aggresive, audibly clipping at some points. At the moment I use a Evox (Siemens) MMKO type of 1uF. This gives a comparible sound as the Wima, maybe a bit better.The value of this cap isn't very critical.
Also the type/brand of output transistors influence the sound, several types have been proposed for this (check the forum at www.diyaudio.com ). When Rudy and I did a comparison between his and mine amp, there was a clear difference. This was mainly because of the bad input caps I had used, but a suspectance arose that the brand of the 2N30555's had something to do with it. Rudy used (old) Motorola's and I used the readily available SGS-Thomson's. Rudy also tried the ST's and said that the Motorola's sounded better! Tim Andrew pointed out to me that the Q with the highest gain factor should be next to the 2N1711. This can make distortion go down by 90% according to Tim. I didn't try this, but will do it when I install other transistors.
ere you can see the modified schematics, I kept the part numbers the same as in the original (assymetrical) schematic.
Note: R3 can be build in for quiscent monitoring purposes but it is advised to bride it when no monitoring is done. R12 has become 68k in stead of the 100k in the original circuit. C5 is polarised oposite incomparison with the '69 circuit!
Switch on blubs...
The '69 version of the JLH had an elco in the output line so the speakers could be directly connected to this elco, if the amp was decently build (all the supply elco bridged with small capacitors) no switch-on-blub prevention was needed. However the Evo versions output is DC coupled so all the rubbish from the supply can be send directly through your expensive speakers.
When the Evo is switched on, the loudspeaker output jumps to 0.8 Volt, and then slowly decreases to about zero volt (in my case some 10 milivolts). So it is wise to protect your speaker with a switch on blub eliminator circuit (!). I got such a thing from an issue of the dutch magazine Elektuur, I adapted it a bit.
The contacts of the drawn relays are placed between the amplifier output and your speaker, there are two relays, one per channel. I used a 24V relay, so the transformer voltage is adapted for it. The resistor R16 determines how much of the rectified voltage is put on the relays, so you should adapt this for your relay. Example;
My relays were 24V types, at a current of 50mA. So R16 should also pass 50mA, and it must eat 4 Volts, since the voltage from my rectifier was 28 Volts. Well, R=U/I --> 4/0.05 makes 80 Ohms (82 in E-12 series).
The switch on delay is determined by R17 and C8. In this configuration it takes about 5 seconds, enough for the ouput voltage to stabilize at a low DC value. I made it a stand alone unit, but you can choose to run it from the amplifier supply.
Later on, I visited Rudy who preferred to run this amp without the switch-on blub eliminator. He thinks it is no problem, when the amp is switched on the woofer on the speaker does a few 'strectches' (hardly audible) and the amp is ready to go. I don't know if my amp could also do without this circuit, but I went through the trouble of making it so I wast to use it too!
The real thing
If you would like to see what mine Evo looks like, click here, for Rudy's click here.