Being a good buddy as he is, Jacco shared some of his 2sk135s with me, and not having the counter-part (2sj50) I decided to use them for a nice circlotron amp.
Different topologies were tried and here is what gave the best sonic result - Source Follower ie. power buffer driven with opa1632. I really love this chip (it's not an OpAmp but a full differential amp) - it sounds most natural and most appealing and I didn't found better sounding circuit as long as you don't ask for more than 12-13Vpeak per phase (per output) which is all I needed to drive this circlotron to full power.
I used two pairs of MOSFETs per channel 'cause I want them to drive 4 Ohm loads easily without worries about Pd.
The amp is very stable (thermally too) - DC offset drifts 2mV from 20 to 60 degrees Celsius on heatsink. Offset to GND is +/- 3mV in the same temp. range and the total bias drift is 20mA for 1.2A bias (600mA per MOSFET). BTW, Jacco sent me pretty well matched MOSFETs (8 pieces are in 50mV range Vgs difference).
With not so good matching Rsource should be 0R47. Off course, 2sk1058 can be used instead of 2sk135.
R29 and R30 - 100k multi-turn pots are used to set the MOSFET's Vgs => Id => DC offset in pretty much the same way as in F5.
The sound is very rewarding, and since this is my first circlotron now I understand why Broskie, EUVL, Circlomanen and some other guys often use this topology.
Different topologies were tried and here is what gave the best sonic result - Source Follower ie. power buffer driven with opa1632. I really love this chip (it's not an OpAmp but a full differential amp) - it sounds most natural and most appealing and I didn't found better sounding circuit as long as you don't ask for more than 12-13Vpeak per phase (per output) which is all I needed to drive this circlotron to full power.
I used two pairs of MOSFETs per channel 'cause I want them to drive 4 Ohm loads easily without worries about Pd.
The amp is very stable (thermally too) - DC offset drifts 2mV from 20 to 60 degrees Celsius on heatsink. Offset to GND is +/- 3mV in the same temp. range and the total bias drift is 20mA for 1.2A bias (600mA per MOSFET). BTW, Jacco sent me pretty well matched MOSFETs (8 pieces are in 50mV range Vgs difference).
With not so good matching Rsource should be 0R47. Off course, 2sk1058 can be used instead of 2sk135.
R29 and R30 - 100k multi-turn pots are used to set the MOSFET's Vgs => Id => DC offset in pretty much the same way as in F5.
The sound is very rewarding, and since this is my first circlotron now I understand why Broskie, EUVL, Circlomanen and some other guys often use this topology.
Attachments
touché Jacco & Juma. It's one of my old favorites too! Now we just need to convince N.P. to work his magic on one of those variants too!
Hey Juma,
Even TI says OP-amp about the 78dB open loop gain, "Black Widow" in the middle of your PCB. The topology of the OP is like a typical poweramp with folded-cascode, except for its separate output-buffers with some kind of error amp(?). But it might sound OK anyway😉.
Still the 2SK135(I have bunch too) output stage looks simple and straightforward.
Hi revintage,
that's the thing in audio, quite often you won't know unless you build it 😉
How is the progress with your gain stage (JFET/Gyrator) ?
Hello!
Getting a circlotron running (and correctly biasing the output stage) is not a very easy task. I like the use of capacitors and the special opamp that also has an inverted output. However, isn't it a problem when the output swings to positive (opamp output 4 is positive), the bias of the negative (lower) side is turned off since the output 5 is negative and the sources of both output halves are referenced to the ground through R5, R4 and R17, R8? Won't this cause some distortion that does not happen in a conventional push-pull output stage? This might not be a problem because circlotrons usually sound very good 🙂 I have built about 4-5 such amps but with a bit different schematics.
Getting a circlotron running (and correctly biasing the output stage) is not a very easy task. I like the use of capacitors and the special opamp that also has an inverted output. However, isn't it a problem when the output swings to positive (opamp output 4 is positive), the bias of the negative (lower) side is turned off since the output 5 is negative and the sources of both output halves are referenced to the ground through R5, R4 and R17, R8? Won't this cause some distortion that does not happen in a conventional push-pull output stage? This might not be a problem because circlotrons usually sound very good 🙂 I have built about 4-5 such amps but with a bit different schematics.
Going slow as always. Spent more time with my woman, wine and whiskey than expected during the holidays 😎.
But got the sand(2SK223, 2SK170 and IRF610) I ordered anyway. So if I could stay away from the above mentioned it could soon be done.
Sure, we'll wait for Nelson to find a perfect active device, then it won't be hard to make it less overcomplicated
This amp (circlotron) found a new home before I finished F5 with 2SK2013/2SJ313 so I didn't have a chance to compare them directly - that's why I can't provide a fair comparison.
Fair enough; but from subjective pleasure, provided the rest of the system was the same, what would you say?
I am asking because I have all the parts except for the THS4131/OPA1632 which could easily be obtained from Farnell or RS. I want to drive a pair of Fostex 168SigmaES drivers in Mass Loaded Qaurter Wave Tubes. Thanks again.
I am asking because I have all the parts except for the THS4131/OPA1632 which could easily be obtained from Farnell or RS. I want to drive a pair of Fostex 168SigmaES drivers in Mass Loaded Qaurter Wave Tubes. Thanks again.
Circlotron asks for precise and neat balancing, and it takes 4 x 25V and +/-15V PSU. OTOH, the F5 with k2013/j313 is less demanding build by far, and both are great amps, so it's up to you to pick your favorite - as the saying goes: build both to be 100% sure what you really want...
Juma: Have you had the opportunity to measure THD on your circlotron? How would you compare your implementation to that of Micheal Rothacher? http://www.diyaudio.com/forums/pass-labs/130810-amazing-fet-circlotron-mike-rothacher.html?highlight=circlotron
No
They are different amps - Michael extended F5 idea to circlotron topology, while I used Laterals as Source Followers in my circlotron amp. I tested both and built what I liked better. You might prefer Michael's amp - a matter of subjective perception...
Juma,
in this post http://www.diyaudio.com/forums/pass...circlotron-mike-rothacher-17.html#post2551991 you replied to Massimo about the possibility of use of the frontend as a standalone SE-to-balanced pre:
As I'm interested in inserting before my (balanced) pre (Pumpkin), I intend to avoid any gain.
Does it simply means that the couples R20/R19 and R16/R21 have to be identical (let's say 3,3K to mantain the same Rin)?
What is the best value, if different?
Many thanks
GuidoR
in this post http://www.diyaudio.com/forums/pass...circlotron-mike-rothacher-17.html#post2551991 you replied to Massimo about the possibility of use of the frontend as a standalone SE-to-balanced pre:
As I'm interested in inserting before my (balanced) pre (Pumpkin), I intend to avoid any gain.
Does it simply means that the couples R20/R19 and R16/R21 have to be identical (let's say 3,3K to mantain the same Rin)?
What is the best value, if different?
Many thanks
GuidoR
Yes
The best value would be the lowest one that can be easily driven by previous stage (because it is inverting configuration and Zin is mainly determined by R16).
I strongly suggest you to read carefully the ths4131 datasheet.
- Status
- Not open for further replies.