FYI, the Darkvoice 336se is a popular tube headphone amp from China. It can roughly be compared to Bottlehead Crack, the latter being better built and pricier. I have not made such comparisons, however.
A few mods have been suggested for the 336se.
I did the Preamp Output ground (remove) mod. I think that made a positive improvement.
Next, I did the LED mod (replacing the 1k R). I used std. red LEDs. Not sure about the results. This may have worsened the sound???? Not sure!
Have there been comments on the brand or type of LED to use? I've heard that color can make a diff. (because diff. colors have different doping materials and diff. current requirements).
Other than tube rolling -- unless you know of any real juicy tubes to install -- suggest whatever comes to mind for further improvement of stock unit.
References:
Darkvoice and Crack Mods | Super Best Audio Friends
Massdrop--thread on 336se
Darkvoice 336i & 336SE Tuberolling PartII | Page 65 | Headphone Reviews and Discussion - Head-Fi.org
Last edited:
May I see the rest of the schematic, for the power supply, as originally used? Or, at least, the value of B+ voltage?
I intend to replicate this schematic to test the sound...I had previously not so great experience with OTL on headphone, but different schematic and tubes. Also different - worse - headphones comparing to present days...Sennheiser HD650 or 560S being my favorites...
I intend to replicate this schematic to test the sound...I had previously not so great experience with OTL on headphone, but different schematic and tubes. Also different - worse - headphones comparing to present days...Sennheiser HD650 or 560S being my favorites...
First I'm going to ask the obvious existential question - what is better?
* bandwidth?
* noise floor?
* transient response?
* frequency response with your chosen headphones?
Big filter caps in there as a low pass filter protecting the front end 6SN7 for noise - options here are
* change caps to a choke+caps (essentially changing from the bi-directional CR low pass filter)
* more capacitance on the LPF if it reduces the cutoff point (very careful of inrush current which will damage the transformer).
You can also do that for the entire supply - but again, it would need design work to ensure you understand if the power transformer could handle it.
Other options:
* Constant Current Source (CCS) to flatten the operating line.
* move power switching to the back so the power is not routed up next to the preamp section (this could be a power relay operated by a simple transistor with the transistor controlled by a low power DC switch on the front instead).
* low noise voltage regulation (likely to need a different power transformer due to the voltage drop)
* replace power supply completely
* buy two dark voices and *with design work* combine the two into a bi-monoblock this then allows the tubes to be paralleled for a single channel..
* differential amplification..
* cascode..
* adding negative feedback..
However we're at the point of a new amp.
As carageae has said - what headphones are you using with it? Low impedance/high impedance?
Also what sources are you using?
Also were the parts provided? You could change the 10uF caps to something better (I don't mean more expensive). The same voltage rating and capacitance but with better current handling. Also 30uF is quite small as this impacts the frequency response as a high pass filter so it would be interesting to see a measured frequency response. Again this would need design work to get the best out of a single triode.
I see they state change the caps for MKP - metalised plastic film - which is all good but you could go one better and simply go straight to metal foil (ie FKP) instead. Careful of the dimensions as foil caps are typically physically larger than MKP. At this point you may end up changing the case.
Without going into the operating points etc and using an oscilloscope, there's very few magical changes on a simple design. We don't know the operating points or the resulting output of harmonics etc.
* bandwidth?
* noise floor?
* transient response?
* frequency response with your chosen headphones?
Big filter caps in there as a low pass filter protecting the front end 6SN7 for noise - options here are
* change caps to a choke+caps (essentially changing from the bi-directional CR low pass filter)
* more capacitance on the LPF if it reduces the cutoff point (very careful of inrush current which will damage the transformer).
You can also do that for the entire supply - but again, it would need design work to ensure you understand if the power transformer could handle it.
Other options:
* Constant Current Source (CCS) to flatten the operating line.
* move power switching to the back so the power is not routed up next to the preamp section (this could be a power relay operated by a simple transistor with the transistor controlled by a low power DC switch on the front instead).
* low noise voltage regulation (likely to need a different power transformer due to the voltage drop)
* replace power supply completely
* buy two dark voices and *with design work* combine the two into a bi-monoblock this then allows the tubes to be paralleled for a single channel..
* differential amplification..
* cascode..
* adding negative feedback..
However we're at the point of a new amp.
As carageae has said - what headphones are you using with it? Low impedance/high impedance?
Also what sources are you using?
Also were the parts provided? You could change the 10uF caps to something better (I don't mean more expensive). The same voltage rating and capacitance but with better current handling. Also 30uF is quite small as this impacts the frequency response as a high pass filter so it would be interesting to see a measured frequency response. Again this would need design work to get the best out of a single triode.
I see they state change the caps for MKP - metalised plastic film - which is all good but you could go one better and simply go straight to metal foil (ie FKP) instead. Careful of the dimensions as foil caps are typically physically larger than MKP. At this point you may end up changing the case.
Without going into the operating points etc and using an oscilloscope, there's very few magical changes on a simple design. We don't know the operating points or the resulting output of harmonics etc.
Last edited:
Just a point on output caps and low impedance.
For 32ohm, the output capacitance of 2200uF is a good ball park. However. That is absolutely massive amount for foil caps and even MKP. This is why you see bi-polar electrolytics being used - I see someone on one of the links has used a mundorf for example.
The problem with electrolytics in output caps are:
* bi-polar should be used - although your design is class A should sit +ve on startup, shutdown and DC offset, other designs of output may not.
* most bi-polars stop at 100V rating unless they're specialist (read expensive).
* high ESR comparatively means running more in parallel (perhaps more than are needed), this is like putting a resistor inline with your output, so parallel caps results in parallel resistances that overall reduces the total resistance.
* they degrade over time- at which point they leak DC current and expose your headphones to DC (bad).
They're not all bad. My OTL headphone amp build is very different but uses MUSE ES 330uF caps - 4x4 for each channel (each channel has 4 output triode pairs) with each pair also having a 0.22uF FKP1 bypass. A total of over 5,000uF per channel is planned but I may drop that to about 3K. However I would not advise this without working through the design of your amp, including the startup/shutdown and operation on a variac and current limiter.
For 32ohm, the output capacitance of 2200uF is a good ball park. However. That is absolutely massive amount for foil caps and even MKP. This is why you see bi-polar electrolytics being used - I see someone on one of the links has used a mundorf for example.
The problem with electrolytics in output caps are:
* bi-polar should be used - although your design is class A should sit +ve on startup, shutdown and DC offset, other designs of output may not.
* most bi-polars stop at 100V rating unless they're specialist (read expensive).
* high ESR comparatively means running more in parallel (perhaps more than are needed), this is like putting a resistor inline with your output, so parallel caps results in parallel resistances that overall reduces the total resistance.
* they degrade over time- at which point they leak DC current and expose your headphones to DC (bad).
They're not all bad. My OTL headphone amp build is very different but uses MUSE ES 330uF caps - 4x4 for each channel (each channel has 4 output triode pairs) with each pair also having a 0.22uF FKP1 bypass. A total of over 5,000uF per channel is planned but I may drop that to about 3K. However I would not advise this without working through the design of your amp, including the startup/shutdown and operation on a variac and current limiter.
OK...interesting points of view....but I'm still waiting for an answer...it seems the initial post belongs to a banned user, so I'm asking anyone who has or has had the hands on this amp: what is the original supply voltage?
I did not calculate the optimum value for the output capacitance, but If the voltage is not so high, two qualitiy electrolitic of reasonable capacity series conected should do the job...30 uF seems a little small value for 32 ohms....the advantage of using unpolarized capacitors should be surpassed by the low frequency cutoff...but using 150 or 300 ohms headphones should have enough low frequency response...still, I doubt that for 32 ohms headphones is 30 uF a plentifull value...I intend using two 470uF electrolytics per channel...235uF should be enough also for 32 ohms headphones...and the safety is reasonable using a pair of 250V+ electrolytics for, let's assume maximum 200V supply.
I did not calculate the optimum value for the output capacitance, but If the voltage is not so high, two qualitiy electrolitic of reasonable capacity series conected should do the job...30 uF seems a little small value for 32 ohms....the advantage of using unpolarized capacitors should be surpassed by the low frequency cutoff...but using 150 or 300 ohms headphones should have enough low frequency response...still, I doubt that for 32 ohms headphones is 30 uF a plentifull value...I intend using two 470uF electrolytics per channel...235uF should be enough also for 32 ohms headphones...and the safety is reasonable using a pair of 250V+ electrolytics for, let's assume maximum 200V supply.
Last edited:
As long as the caps are rated above the rectified voltage - calculated at the upper limit margin of the mains and the transformer regulation. Here in the UK, we get ±10% mains for example. It's quoted as 230V ±10%. I've measured mine at 246V consistently. If the DV is full bridge rectified unregulated then my voltage is 246*1.414 = 347Vdc. The max needs 375Vdc not including any minor AC transient bumps. So 450V tends to be normal ratings. (This is more for any DV reader rather than carageae whom I suspect knows this already!)
Given a shorted tube, the output cap then has full B+ across it. Naturally you'll need to calculated that for your own country's requirements.
For applications where the DC polarity (including any AC down swing) doesn't reverse then a non-polar should be fine.
For the low pass filter this is a good tool: (Sample)RC Low-pass Filter Design Tool - Result -
I'd still look at having smaller capacitance caps in parallel due to the series resistance - say three 150uF 450V 105degC caps. If your voltages are lower that's a bonus (in physical size). Smaller electrolytics simply have a higher ESR. The current DV uses what appear to be sprague orange drops (or copies) in parallel which implies metalised plastic or foil. So simply replacing them is likely to sacrifice some clarity for bass response. So you may want a foil over the top of the electrolytic.
Next you'll be buying all sorts of caps then mix and matching to get the sound you like 😀
Given a shorted tube, the output cap then has full B+ across it. Naturally you'll need to calculated that for your own country's requirements.
For applications where the DC polarity (including any AC down swing) doesn't reverse then a non-polar should be fine.
For the low pass filter this is a good tool: (Sample)RC Low-pass Filter Design Tool - Result -
I'd still look at having smaller capacitance caps in parallel due to the series resistance - say three 150uF 450V 105degC caps. If your voltages are lower that's a bonus (in physical size). Smaller electrolytics simply have a higher ESR. The current DV uses what appear to be sprague orange drops (or copies) in parallel which implies metalised plastic or foil. So simply replacing them is likely to sacrifice some clarity for bass response. So you may want a foil over the top of the electrolytic.
Next you'll be buying all sorts of caps then mix and matching to get the sound you like 😀
Last edited:
John Broskie had an article about how he would modify a Darkvoice may be of interest...
Cathode-Follower OTL and SIT FETs
Cathode-Follower OTL and SIT FETs
What about this mod?
I wait for friends knowing LT Spice to complete the schematic...I have the tubes, and from what I read, it's too weak the 25W6GT for normal SE amplifier. But should be plentifull for headphone amp.
Darkvoice ETC mod
I wait for friends knowing LT Spice to complete the schematic...I have the tubes, and from what I read, it's too weak the 25W6GT for normal SE amplifier. But should be plentifull for headphone amp.
Darkvoice ETC mod
Last edited:
For which tube? (edit: I figured out you mean the voltage swing from the 6sn7)
That 25W6GT is a 60mA, 8000 transconductance but a 13Kohm plate resistance tube. Compare that to a 6AS7 and 6080 with 280ohm plate resistance with 7000 tansconductance.
That's quite a large plate resistance if you're running as cathode follower? An ecc99 has a 2.3Kohm plate resistance for example. You may find that 13Kohm increases the output impedance?
I see you've switched the input tube to ECC83 ie a 12ax7. So the front end has now moved from a 20mu to 100mu..
So I take it that you're switching out the more current focus to voltage swing - ie large swing 12ax7 into a 25W6GT for a larger voltage swing output into high impedance headphones?
(sorry - writing as I'm thinking about it)
Last edited:
It's ECC82 or the octal Russian 6n8s,so there's no talk about higher mu ECC83. Eventually I could use an ECC85, for a personal reason - I have plenty of UCC85, fits ok along with the 25V heater final tubes. But I think it's enough gain with ECC82...I will change only if it's a better operating point for the following tube.
I know the chosen pentode has much higher impedance....but I was thinking the impedance is enough regarding the small power needed for 150 ohms headphone...but more suggestions are welcome.
If the plate impedance is such an issue, I could use two russian 6N6, 2 parallel triodes per channel. The 6080 amplifier claims to be ok using 32 ohms headphones, I intend not to use tubes for them....I will try this schematic with 80, 150 and 300 ohm headphones, which I think they are suitable for this amp.
I know the chosen pentode has much higher impedance....but I was thinking the impedance is enough regarding the small power needed for 150 ohms headphone...but more suggestions are welcome.
If the plate impedance is such an issue, I could use two russian 6N6, 2 parallel triodes per channel. The 6080 amplifier claims to be ok using 32 ohms headphones, I intend not to use tubes for them....I will try this schematic with 80, 150 and 300 ohm headphones, which I think they are suitable for this amp.
Last edited: