Has anyone ever cloned a Leak Stereo 20? I've been looking to build an easy stereo push-pull amp in the 10-20 watt range using Hammond iron for OTs and everything I have read about vintage Leak gear has been positive. I originally wanted to clone a Quad II or Leak TL/12.1 but decided against monoblocks because I don't want to buy two power transformers, and because KT66s are expensive! I am looking to use the Hammon 1609(10k primary), 1615(5k) or 1620(6.6k) output transformer as they are all very close in terms of price, hopefully they sound good too.
Would a Stereo 20 be a good choice or should I go with something else? There does not appear to be a choke in the power supply(unlike the monoblock TL/12+) but I think I would like to use one in place of the first resistor in the PS if there is likely to be a significant impact on ripple. I've also thought about using 6V6s instead of EL84.
here is a schematic for the Stereo 20
http://ampslab.com/SCHEMATICS/LeakStereo20.gif
also, i'm interested in this Mullard EL34 schematic as an alternative.
http://bonavolta.ch/hobby/en/audio/el34_1.htm
thanks
Would a Stereo 20 be a good choice or should I go with something else? There does not appear to be a choke in the power supply(unlike the monoblock TL/12+) but I think I would like to use one in place of the first resistor in the PS if there is likely to be a significant impact on ripple. I've also thought about using 6V6s instead of EL84.
here is a schematic for the Stereo 20
http://ampslab.com/SCHEMATICS/LeakStereo20.gif
also, i'm interested in this Mullard EL34 schematic as an alternative.
http://bonavolta.ch/hobby/en/audio/el34_1.htm
thanks
The Stereo 20 (like all Leak amplifiers) is ludicrously over-sensitive at 120mV for full power. To get that gain, it uses more coupling capacitors (thereby compromising LF stability) than is ideal. It was a good design for its time and sounded better than a Quad II, but without identical output transformers you'd be hard-pushed to equal its performance, let alone match a latter-day design that doesn't try to screw every last dB of gain out of the valves.
hi, l'm no expert but l agree with another poster who says forget it, the sound of these classic amps depended bon the o/p trannies more than anything modern equivelants will more than likely not give you anything like the sound ofthe originals, if you want somethig to build that sounds terrifi try a pp 6aq5 amp the 6aq5 is a6v6 in a miniature envelope, from all accounts they make great sounding amp do a search on audio asylum for info on them...... cheers
TC
Yes, I 'cloned' one half of a Leak Stereo 20 back in 1962. I didn't want two channels, because mono was still the mainstream for radio and records in those days. Anyway, as a schoolboy, I didn't have the money for two OP transformers. I chose a high quality OPT, though, one made by Parmeko for the Mullard 5-10. I also built one half of the Leak Varislope preamp, which I simplified to meet my needs. It all seemed to work OK, at least to my inexperienced ears.
However, the Leak Stereo 20 has a number of design weaknesses, as I've come to realize since then, including:
* AC coupling between the first stage and the splitter, which can cause LF instability in the negative feedback loop, particularly if you use a rather ordinary OP transformer.
* the use of 12AX7 for both the first (voltage amp) and second (splitter/driver) stages; it's inclined to accentuate the non-linearities that are peculiar to that particular tube type, which is not all that linear to start with, and it might have sounded better if Leak had chosen a different, preferably lower impedance tube for the splitter/driver .
* no constant current sink (CCS) in the tail of the LTP splitter, which would be very much better than the resistor Leak used.
* the use of a resistor instead of a choke for PS smoothing, weakening the bass response and causing higher ripple.
* very heavy negative feedback, which is excessive for an ultralinear design (they believed that there was no such thing as too much NFB in those days).
* high input sensitivity.
In the 1960s, the public was not very discriminating and something like the Leak Stereo 20 was many orders of magnitude better than the run-of-the-mill radio sets, record players and radiograms on the market at that time.
Leak was cost-cutting when he decided to use a smoothing resistor instead of a choke. You are only building one amp for personal use, not to make a profit, so you don't need to make such 'economies' and a choke would be a much better idea.
I prefer the Mullard 5-10 or 5-20, which is the other alternative you mention, but it shares many of the weaknesses of the Leak Stereo 20. Its open loop gain, negative feedback and input sensitivity are all excessive - even worse than Leak's! But at least the Mullard circuit has direct coupling between the first and second stages, so it is inherently more stable.
I believe the Mullard 5-10 or 5-20 would be a better place to start than the Leak, but it should be improved by fixing all the weaknesses it has in common with the Stereo 20, including using a choke in the power supply and a lower impedance medium mu double triode as the splitter/driver (e.g. 6CG7 or 6FQ7). This would give excellent balance, despite its lower mu, if you used a CCS instead of a resistor in the tail. I would also omit the cathode bypass cap from the EF86 but I wouldn't triode-strap it, as some people do, because the gain would already be reduced a lot by using a lower mu splitter tube.
I wouldn't advise you to use a Hammond OP transformer, since this is an ultralinear design and requires a special quality OPT to work well. Hammonds are not bad but IMHO there are better transformers for your purposes.
However, the Leak Stereo 20 has a number of design weaknesses, as I've come to realize since then, including:
* AC coupling between the first stage and the splitter, which can cause LF instability in the negative feedback loop, particularly if you use a rather ordinary OP transformer.
* the use of 12AX7 for both the first (voltage amp) and second (splitter/driver) stages; it's inclined to accentuate the non-linearities that are peculiar to that particular tube type, which is not all that linear to start with, and it might have sounded better if Leak had chosen a different, preferably lower impedance tube for the splitter/driver .
* no constant current sink (CCS) in the tail of the LTP splitter, which would be very much better than the resistor Leak used.
* the use of a resistor instead of a choke for PS smoothing, weakening the bass response and causing higher ripple.
* very heavy negative feedback, which is excessive for an ultralinear design (they believed that there was no such thing as too much NFB in those days).
* high input sensitivity.
In the 1960s, the public was not very discriminating and something like the Leak Stereo 20 was many orders of magnitude better than the run-of-the-mill radio sets, record players and radiograms on the market at that time.
Leak was cost-cutting when he decided to use a smoothing resistor instead of a choke. You are only building one amp for personal use, not to make a profit, so you don't need to make such 'economies' and a choke would be a much better idea.
I prefer the Mullard 5-10 or 5-20, which is the other alternative you mention, but it shares many of the weaknesses of the Leak Stereo 20. Its open loop gain, negative feedback and input sensitivity are all excessive - even worse than Leak's! But at least the Mullard circuit has direct coupling between the first and second stages, so it is inherently more stable.
I believe the Mullard 5-10 or 5-20 would be a better place to start than the Leak, but it should be improved by fixing all the weaknesses it has in common with the Stereo 20, including using a choke in the power supply and a lower impedance medium mu double triode as the splitter/driver (e.g. 6CG7 or 6FQ7). This would give excellent balance, despite its lower mu, if you used a CCS instead of a resistor in the tail. I would also omit the cathode bypass cap from the EF86 but I wouldn't triode-strap it, as some people do, because the gain would already be reduced a lot by using a lower mu splitter tube.
I wouldn't advise you to use a Hammond OP transformer, since this is an ultralinear design and requires a special quality OPT to work well. Hammonds are not bad but IMHO there are better transformers for your purposes.
Bikehorn,
Hammond 1650Es will work in an UltraLinear wired "El Cheapo". The 'AQ5 called for is a 6V6 in a 7 pin mini package. IMO, the DynaClone Z565 is better than Hammond "iron", but it's all too easy to spend somebody else's money.
Hammond 1650Es will work in an UltraLinear wired "El Cheapo". The 'AQ5 called for is a 6V6 in a 7 pin mini package.
IMO, the DynaClone Z565 is better than Hammond "iron", but it's all too easy to spend somebody else's money.I've probably made at least 30 different versions of the Stereo 20, and I'm still not satisfied! I agree with the others - build something else. I'd suggest Allen Wright's PP1 amp on his website www.vacuumstate.com in the schematics. It's simple, effective, and will teach you some useful techniques like constant current sinks and having a balanced input option. You can go on and change the output stage to other valves like 2a3, 6B4G, 300b at a later date since the input stage is quite versatile.
Well I guess the Stereo 20's out...
I've definitely been interested in the 6AQ5, in fact if i had built the Leak with 6V6s I probably would have gone with those. NOS 6AQ5s sell for way less than new production 6V6s. I also thought about building a Dynaco ST35 clone per DIYtube but the available clone chassis seem to be awkward and clunky
As transformers go, I need something that won't break the bank, or else I'll have to build gainclones. I will not need a lot of power, 15 watts would probably be enough as most of the time I will be using efficient speakers(at least 96-97 dB). Even if they aren't the best, on a relative scale i'm sure they'll handily outperform the mid 70's solid state Sony stereo receiver I have What other options are available for outputs? The Dynaclones look good. Will it be easy to include a triode/pentode switch? can i run the 6AQ5s in the El Cheapo in pentode mode just by connecting the screen grid taps on the transformer? i've seen this feature on some guitar amps...obviously the purpose would be a little different on a hifi amp.
I've definitely been interested in the 6AQ5, in fact if i had built the Leak with 6V6s I probably would have gone with those. NOS 6AQ5s sell for way less than new production 6V6s. I also thought about building a Dynaco ST35 clone per DIYtube but the available clone chassis seem to be awkward and clunky
As transformers go, I need something that won't break the bank, or else I'll have to build gainclones. I will not need a lot of power, 15 watts would probably be enough as most of the time I will be using efficient speakers(at least 96-97 dB). Even if they aren't the best, on a relative scale i'm sure they'll handily outperform the mid 70's solid state Sony stereo receiver I have
What other options are available for outputs? The Dynaclones look good. Will it be easy to include a triode/pentode switch? can i run the 6AQ5s in the El Cheapo in pentode mode just by connecting the screen grid taps on the transformer? i've seen this feature on some guitar amps...obviously the purpose would be a little different on a hifi amp.
You don't need a terribly expensive OP transformer, just a good one! Your power requirements are modest, so cost should be reasonable unless you choose an especially expensive brand. I expect you can get plenty of helpful suggestions from others here.
From a data sheet I've seen, a pair of 6AQ5s in pentode push-pull can give up to 10 watts output (quite respectable for such a tiny tube!) with 10k plate-to-plate (P-P) load. I don't know what load impedance they would need in triode-mode but, in any case, you'd probably find the available power inadequate, maybe 2.5 watts.
If you want to try switching between pentode and triode, it can possibly be done in the way you suggest, but you have to consider the required P-P load in each case. If compromise is necessary, it would be better to base the OP transformer on the needs of pentode mode, since this is quite critical. Triode-mode is less fussy, and is particularly forgiving if the P-P load is a higher impedance than the ideal load - you just get less power and less distortion.
Your power requirements are modest, so cost should be reasonable unless you choose an especially expensive brand. I expect you can get plenty of helpful suggestions from others here.From a data sheet I've seen, a pair of 6AQ5s in pentode push-pull can give up to 10 watts output (quite respectable for such a tiny tube!) with 10k plate-to-plate (P-P) load. I don't know what load impedance they would need in triode-mode but, in any case, you'd probably find the available power inadequate, maybe 2.5 watts.
If you want to try switching between pentode and triode, it can possibly be done in the way you suggest, but you have to consider the required P-P load in each case. If compromise is necessary, it would be better to base the OP transformer on the needs of pentode mode, since this is quite critical. Triode-mode is less fussy, and is particularly forgiving if the P-P load is a higher impedance than the ideal load - you just get less power and less distortion.
bikehorn said:Well I guess the Stereo 20's out...
I've definitely been interested in the 6AQ5, in fact if i had built the Leak with 6V6s I probably would have gone with those. NOS 6AQ5s sell for way less than new production 6V6s. I also thought about building a Dynaco ST35 clone per DIYtube but the available clone chassis seem to be awkward and clunky
As transformers go, I need something that won't break the bank, or else I'll have to build gainclones. I will not need a lot of power, 15 watts would probably be enough as most of the time I will be using efficient speakers(at least 96-97 dB). Even if they aren't the best, on a relative scale i'm sure they'll handily outperform the mid 70's solid state Sony stereo receiver I have
Paul Joppa has presented a 102 dB. rule. In an "average" room, an amp/speaker combo should be capable of 102 dB. SPL peaks at a 1 M. distance. So, 4 WPC is enough with 96 dB. efficient speakers. As another poster indicated, you will be fine with an "El Cheapo" wired strictly triode.
Hammond 1650Es are certainly no worse than the guitar amp trafos spec'd as minimum for an "El Cheapo". Allied Electronics has a Canadian warehouse and they carry both Hammond and Triad. You can bundle your complete "iron" order from Allied, which will save some shipping charges and avoid Customs hassles.
While it's a lot to wade thru, I strongly suggest you read the entire "El Cheapo" thread. You will get a feel for how the design developed and also get builder feedback.
BTW, your mid 70s Sony receiver aint half bad. Japanese consumer electronics made in that time frame is vastly superior to the JUNK currently marketed. Connect the recording O/Ps on the Sony to the I/P jacks on the "El Cheapo". You get access to the tuner, phono section, and line level I/Ps, while bypassing the line stage and power amp sections. If by happy circumstance, the receiver has the preamp-out/amp-in feature, you can use the power amp section to power a subwoofer.
Thanks for all the replies so far guys. the 6AQ5 is actually just the tube i wanted to use.
The drivers I was planning to use are P. Audio BM15CXHBs in Karlson cabinets. these are 15" coaxials with a claimed sensitivity of 101 dB for the woofer and 106 dB for the compression driver. I think the sensitivity, at least for the woofer, has the potential to be a little optimistic although it should still be pretty high. I have also heard that the compression driver's sensitivity is lower than specified, but I have no concrete proof of this so that could be false(someone on audioasylum said that...but who knows). So, according to the driver specs i should need only very little power...but these may not be the only speakers i ever use and if I happen to use something with lower sensitivity in the future I don't want to be caught with just 4 watts. I would hate to not be able to use a great sounding amp that i built myself just because it doesn't have a little muscle in reserve to drive my speakers. This is why I want at least 10 watts.
I can live without triode mode. I know it's supposed to be more linear and all but I'm sure a decent pentode output should be just fine and i'd rather not compromise one mode of operation for the other. Besides, I think upgrading my sources would make a bigger difference in sound. Maybe next time i'll try triodes...with 6C33Cs Always liked the bottle shape, and the output potential!
I've been reading through parts of the El Cheapo thread...and one thing I realised quickly is I don't like the power supply. I want to be able to accomplish everything with one transformer, which I think should be possible with a good one, and I would also like to use a tube rectifier...not for philosophical reasons...I just want to hear what it sounds like. Designing my own PS might be a fun exercise too.
How much benefit would there be to stablized screen supplies?
The drivers I was planning to use are P. Audio BM15CXHBs in Karlson cabinets. these are 15" coaxials with a claimed sensitivity of 101 dB for the woofer and 106 dB for the compression driver. I think the sensitivity, at least for the woofer, has the potential to be a little optimistic although it should still be pretty high. I have also heard that the compression driver's sensitivity is lower than specified, but I have no concrete proof of this so that could be false(someone on audioasylum said that...but who knows). So, according to the driver specs i should need only very little power...but these may not be the only speakers i ever use and if I happen to use something with lower sensitivity in the future I don't want to be caught with just 4 watts. I would hate to not be able to use a great sounding amp that i built myself just because it doesn't have a little muscle in reserve to drive my speakers. This is why I want at least 10 watts.
I can live without triode mode. I know it's supposed to be more linear and all but I'm sure a decent pentode output should be just fine and i'd rather not compromise one mode of operation for the other. Besides, I think upgrading my sources would make a bigger difference in sound. Maybe next time i'll try triodes...with 6C33Cs
Always liked the bottle shape, and the output potential!I've been reading through parts of the El Cheapo thread...and one thing I realised quickly is I don't like the power supply. I want to be able to accomplish everything with one transformer, which I think should be possible with a good one, and I would also like to use a tube rectifier...not for philosophical reasons...I just want to hear what it sounds like. Designing my own PS might be a fun exercise too.
How much benefit would there be to stablized screen supplies?
phn said:
Yes, I only meant I was scrapping the triode/pentode switch and sticking to one mode. I would build something resembling the TL/12 or Quad II(would love to use KT66) but this means doubling the number of PTs and chokes i need...as well as requiring more expensive output tubes.
Re the sony, I have a pile of old solid state stuff, including a Technics (on loan to a friend) that sounds much better than the sony, and a Superscope i performed a PS upgrade on that sounds better than the Technics, unfortunately the left channel keeps dropping in and out, and I suspect a transistor in the preamp is going. it has a much better phono stage than the Sony too, and is a lot quieter plus has more power. it will be my "utility" amp when I have time to sort it out. I definitely agree about modern consumer electronics sucking.
can i double the number of 6AQ5s for more power with only minor circuit tweaks? i reason that if I can get a 20 watt transformer for the same price as a 10 watt transformer, i may as well use as much of its capacity as i can. obviously this might require a bigger PT.
Remember, I said read the entire thread. Thought has been given to tube rectified "El Cheapo" variants. We're ahead of you!
To get what you want, expenses will go up. TANSTAAFL. Build your "El Cheapo" using ultralinear topology. 15 WPC or thereabouts is a reasonable expectation. Order a complete set of ST35 "iron" from Triode Electronics. The cost is 180 USD plus shipping, Customs, and taxes. Look at the ST35 power trafo data sheet. There is enough current in each of the paired filament windings to supply a 12AT7, 2X 6AQ5s, and a 7Z4. Use a 7Z4, with its plates in parallel, at each end of the rectifier winding. CLC filtration will be employed. The exact value of the 1st cap. has to be determined at the bench with the PSU under load. B+ should be over 360 and under 365 V. All that's left is the B- supply, which will require a separate trafo unless you buy VERY costly custom "iron". I recommend the little Allied brand trafo shown in the schematic. Hybrid bridge rectify with 2X UF4007s and a 6AL5. The 6AL5 gets its heater power from the B- trafo.
This is the best design I could find:
http://www.tubebuilder.com/images/schematics/pushpull/el84pp_ex2.gif
If you don't want UL, take a look at this:
http://www.werple.net.au/~kiewavly/Leak/Circuits/TL25A_cct4.gif
You can use any cheap pentode for regulation. Unfortunately, there doesn't seem to be a 1 Meg trim pot with high enough rating. Otherwise that would be ideal. You can also try connecting screen to anode using a 100 Ohm resistor.
This will be a lot more complicated than following the diagram above. By the way, I don't know about the sensitivity of the EL84 amp. But should be lower than Leak.
http://www.tubebuilder.com/images/schematics/pushpull/el84pp_ex2.gif
If you don't want UL, take a look at this:
http://www.werple.net.au/~kiewavly/Leak/Circuits/TL25A_cct4.gif
You can use any cheap pentode for regulation. Unfortunately, there doesn't seem to be a 1 Meg trim pot with high enough rating. Otherwise that would be ideal. You can also try connecting screen to anode using a 100 Ohm resistor.
This will be a lot more complicated than following the diagram above. By the way, I don't know about the sensitivity of the EL84 amp. But should be lower than Leak.
I didn't look that close. The other EL84 designs I found when using the google image search used more complicated designs or stupid ECC83/12AX7.
Now looking, i does look like Leak, but hopefully not as sensitive.
I did find this as well: http://www.tubebuilder.com/images/schematics/pushpull/el84_4.gif
They are, after all, on the same site: http://www.tubebuilder.com/schematic1.html
Now looking, i does look like Leak, but hopefully not as sensitive.
I did find this as well: http://www.tubebuilder.com/images/schematics/pushpull/el84_4.gif
They are, after all, on the same site: http://www.tubebuilder.com/schematic1.html
Arriving at the sermon as the collection is taken ...
I hate saying this, but there were serious shortcomings in the Leak Stereo 20. I cannot fathom why as I am supportive of HJ Leak (met him), but input/driver working points were quite out - not to analyse here further. I refurbished a few and had to do several component value changes to get them right - sorry. (I liked the classic Leak TL12 a lot and am still using that topology in most tube amps made for clients.)
Unfortunately one will not get more than 11W out of 6AQ5s pp - ratings not sufficient. 6V6s will yield about 13-14W, but that is hardly an audible difference. Also, to be careful: The 6V6/6AQ5 requires some 30% more grid signal than EL84; a slight gain reduction.
Perhaps my most important contribution (done before - sorry - but again relevant) is the unsuitable use of a 12AX7 as a driving tube or even an input stage in power amplifiers. The mu/rp is just simply to high for this purpose compared to e.g. 12AT7, ECC88 or many other similar tubes (been mentioned already). This is also where I personally would change the classic 510/520 design. As the gain is already too high, I would go down to 12AT7 or ECC88 equivalents.
Also herecy on my part in the eyes of some (here we go!), but I am uncertain that a CCS in the Schmitt phase inverter gives enough essential advantage over a simple resistor. The impedance at the cathodes of e.g. a 12AT7 moderately hard driven (RL of 68K or lower for 350V ht) is about 2K. Thus any common cathode resistor over 20K is giving diminishing returns. I am not condemning a CCS, only saying that some of the simple designs existing are worse than resistors at high frequency. The thing does need frequency characteristics, etc. (Using a Rk resistor again needs adjustment for fine balance, so...)
Finally Bikehorn (again repeating), the characteristics of UL is close enough to triode operation that I cannot see a reason not to use it (20% - 25% taps fior the 6V6) - straight pentode is certainly not an alternative. Some hate it, but with respect, subjective experiences can differ; they have proved to be hardly a yardstick. (As you mentioned the Quad II, that output topology has been my choice for years but you need a special output transformer, so.)
Regards
I hate saying this, but there were serious shortcomings in the Leak Stereo 20. I cannot fathom why as I am supportive of HJ Leak (met him), but input/driver working points were quite out - not to analyse here further. I refurbished a few and had to do several component value changes to get them right - sorry. (I liked the classic Leak TL12 a lot and am still using that topology in most tube amps made for clients.)
Unfortunately one will not get more than 11W out of 6AQ5s pp - ratings not sufficient. 6V6s will yield about 13-14W, but that is hardly an audible difference. Also, to be careful: The 6V6/6AQ5 requires some 30% more grid signal than EL84; a slight gain reduction.
Perhaps my most important contribution (done before - sorry - but again relevant) is the unsuitable use of a 12AX7 as a driving tube or even an input stage in power amplifiers. The mu/rp is just simply to high for this purpose compared to e.g. 12AT7, ECC88 or many other similar tubes (been mentioned already). This is also where I personally would change the classic 510/520 design. As the gain is already too high, I would go down to 12AT7 or ECC88 equivalents.
Also herecy on my part in the eyes of some (here we go!), but I am uncertain that a CCS in the Schmitt phase inverter gives enough essential advantage over a simple resistor. The impedance at the cathodes of e.g. a 12AT7 moderately hard driven (RL of 68K or lower for 350V ht) is about 2K. Thus any common cathode resistor over 20K is giving diminishing returns. I am not condemning a CCS, only saying that some of the simple designs existing are worse than resistors at high frequency. The thing does need frequency characteristics, etc. (Using a Rk resistor again needs adjustment for fine balance, so...)
Finally Bikehorn (again repeating), the characteristics of UL is close enough to triode operation that I cannot see a reason not to use it (20% - 25% taps fior the 6V6) - straight pentode is certainly not an alternative. Some hate it, but with respect, subjective experiences can differ; they have proved to be hardly a yardstick. (As you mentioned the Quad II, that output topology has been my choice for years but you need a special output transformer, so.)
Regards
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.