The "Lovely Cube" Headphone Amp (Lehmann Black Cube Linear Clone)

  1. francisdemarte
    Thanks for the very detailed report of the LC components! What initially made you suspect of the power supply diodes?
  2. BlaBlaBla
    I think, you can not. At least without proper equipment. Instead of that it is better to invest in good coupling caps and good electrolytes before regulator.
    Well I want to explain something...
    Difference between XXX product and YYY product is only in quality control. You can buy 10 XXX caps, and all of them will last, for example, 6 years and in those 6 years will be within specifications (100% quality). You can buy 10 YYY caps and 4 of them will last 6 years (40%) within specification, 4 will last only 2 years (40%) within spec. and 2 will be out of specifications (which means BAD - 20%) right after production. Only 40% are within specification.
    Production (amount of products) is the same for XXX and for YYY. But XXX passes a lots of controls and its production is in constant upgrade and under supervision, and YYY has no quality control and everything what produces - sells on the market, and final users are people who represent "first quality control after production". XXX product is more expensive than YYY, for 5-10 times because more labor, and products that are returned back (recycled).
    But as we see, on the market is a lot of consumer parts that are replaced after year (because they became obsolete - good or bad it does not matter - cell phones, TV-s etc...)... Parts who will last 6 years are not needed anymore because they are replaced from user itself after 1 year. That means that those 40% within spec. are now 80%. And only one that are out of specification right after production are BAD (20%). And usual buyer will say:"Well, yes. Two caps were bad but, they were very cheap. I bought five more. At the end I am satisfied..." So YYY sells more than ever, and have bigger profit. XXX is struggling in a deficit, regards high quality, because such high quality is not needed anymore.
    It is not needed any more, because of us.
    Because of our demands for new technology. We buy new product every time when something comes out. We are buying because something is NEW, and not because it is QUALITY PRODUCT .
    Buyer is guilty. Not the manufacturer. WE are guilty.
    It will happen, that people who want good parts, will have to disassemble 20 years old amplifiers, TVs, etc, and take parts from there.
    Cheers :)
  3. FlukeII

    It wasn't so much the rectifier diodes on the board, it was the output transistors...the kit supplies black devices with the BD139 marking, PH and 91 presumably the counterfeiters idea of a date code...when I saw PH - only PHILIPS used to use this marking on it's on on TRIACs, SCRs and MOSFETs; with BJT's it always used to be PHxy - xy was the year - at least that's what i remember!
    The trouble is the PHILIPS BD139 and BD140 devices, to the best of my knowledge, were only fabed in the Nederland’s and Germany, then packaged in the Nederland’s and Belgium e.g. the silicon wafer was cut and the transistors glued to the Collector terminal and the molded SOT32 package applied.
    Having worked for PHILIPS at the time in 91 when we were using these parts, they only ever came in a light grey plastic housing with printed BD139 and the date code, say PH91 (yr 91), they were not using laser etching based markings until after the 90’s for devices and never on the BD139 and BD140s only made in EU. (Also, the factory would only do one production run per year and put excess into inventory - if you neded more, you had to wait to next year's production run - these were the days before the internet -- life was easy!!)
    (When the ELCOMA section was broken up within PHILIPS and they formed NXP, they kept the production of the BD140 and BD139 devices the same - according to the local distributor)
    So the dead give away was the packaging of these output transistors - they just don't look right.
    Attached is a photo of genuine PHILIPS devices, made in 1995 and 2001 of the BD139 and BD140 (I can’t show a photo of the BD139-16 or BD140-14 as the AMP is at a friends house and a I can’t find the last 10 pcs of each in the spare room – it’s a mess with boxes everywhere…) PH-BD139-BD140.jpg
    For the BD139-16 / BD140-16 the markings are identical except that there is the number 16 placed near the BD139/BD140 text, the same font size. e.g.BD139 16 or BD13916 - depending where the devices were packaged.
    That wasn't enough I thought, so at work we tested the fT and hFE of these parts (supplied with the Lovely Cube kit), and while close to the ST parts performance, they were nothing like the original PHILIPS parts, therefore they can only be counterfeit IMO?
    I suspect the kids putting these kits together were at primary school back in 91 when PHILIPS were making the BD139 and BD140s - so I suspect they have no idea what they are buying from their component suppliers, else they know exactly what they are doing and are too preoccupied chasing Deng Xiaopingi's vision: "'s glorious to be rich..."!
    PS. I would be really amazed if the counterfeiters can clone the original PHILIPS / NXP packaging material (epoxy resin), at PHILIPS they used to make their own resins from raw materials - that's why I'm sure the black BD139s and BD140s supplied in the kit are fake....
  4. BlaBlaBla
    OMFG... After this, I realize that 95% of electronic parts are not genuine. I remember the gray transistors. I think that last time, when I saw them, was in old Sabba TV (20 years old)! So if other parts are equally old (Toshiba) then we can expect nothing....
    Conclusion is: If someone is in the DIY bussiness, top priority is assembling amplifier that works properly (eg. does not make strange noises, hisses or blows headphones), and has parts that have passed basic quality control. We can forget about rest ("only genuine parts"). Only hope is to catch part that is within spec.
    I was worried until now, but too much concern about technical stuff can spoil the enjoyment in music, so:
    - headphones are working properly without cracks and hiss,
    - DC offset is below 3mV on both channels,
    - and amplifier is not overheating.
    - my ears are satisfied,
    Fair enough.
  5. FlukeII
    For the next installment, the Lovely Cube Kit comes with some electrolytics that don't look as though they are the original thing, the clues here are the quality of the shrink wrap labels around the capacitor's cans - it's possible they are genuine, but nonetheless I figured you could do better...
    For that matter, the copper pour of the Lovely Cube PCB doesn't look as heavy as the original Lehmann PCB, which looks as though it's using 2,0 oz copper. The  Lovely Cube PCB looks as though it's using 1,0 oz copper or less?
    For the main smoothing capacitors, the Lehmann Black Cube uses VISHAY / BC 4700uF 40V +105deg C, part number 058 57472, 30mm x 30mm case with a Ir(100Hz)=2,96A, 3,61A and Z(100Hz)=59m Ohm and 40m Ohm @ 10kHz.
    This part is not bad and is easily obtainable from the like of RS Components and Farnell.
    We thought about trying other capacitors in this position, but we found the biggest sonic signature change came with changing the 470uF 25V Electrolytics decoupling the op amp and the diamond buffer stages.
    The original Lehmann used 470uF 25V United Chemicon LXZ - this part seems to be LXZ25VB471M10X16LL, 10x16mm. The specs on this part are pretty good for a switchmode supply type application: Zmax(100kHz)=68m Ohm, and Ir(100kHz)=1,050A, but at audio frequencies (1kHz), you only get about 85%.UPDATE:  The LXZ caps have a different part number at Digi-Key, they stock the ELXZ250ELL471MJ16S 470uF 25V Part#: 565-1958-ND, they say it is the same part but, different part number; no explanation why.
    Later I'll post up a table with a bunch of capacitors we tried, but to cut to the chase the best part that worked here for me was the PANASONIC FM Series (EEUFM1E821L) 820uF 25V (Digi-Key: EEU-FM1E821LB-ND) - the issue is with the capacitors so close to the heatsinks you're limited to about 10mm diam. These were decoupled with 220nF EPCOS Stacked Film MKT capacitors B32520C224J - (Mouser Part#: 871-B32520C224J) these had to be soldered underneath the PCB! These capacitors are in the output stages. This gave a fast tight presentation that wasn't in your face - mids not as warm as when using Nichicon PM Series - Extremely Low Impedance, High Reliability, 680uF 25V (UPM1E681MPD) (Mouser Part# 647-UPM1E681MPD), but it's most pleasant when listening on my T1's, nice Bass on DT770's and fast with the AKG K271 MkII's.
    More to come...[​IMG]
  6. ringer

    What do you think about bipolar Mundorf E-cap 470uF?
  7. BlaBlaBla
    Is not 820uF too much?
    In the specification for LM317/337, they suggest that output cap is between 100uF and 1000uF. Two 820uF (per channel) are 1600uF.
    What about low impedance headphones? Do they need big caps on the output?
  8. ringer


    In my first DIY I use 16000 uF in output of LM317. So I think 1600 uF is OK :)
  9. Knusperfisch
    Well I don't care if these parts are counterfeit products as long as the amp sounds good. I think the slogan "you get what you pay for" fits good in here. You should always consider that the seller wants to make profit, and you should also calculate that you could get fake parts especially if you're buying from China. But I really wonder why the Matrix M-Stage is considered to sound worse in comparison to LC. Does it mean that the M-Stage uses also counterfeit parts ?
    Maybe these transistors are just fake labeled if there on par with the STMicro BD139, BD140? Can't imagine that chinese fabs can produce transistors at STMicro quality which is definitaly not the worst in the industry.
    All I can say is that as long as the quality of the parts don't deceed a minimum level, sound good and live long I can live with ungenuine transistors. And if I want to upgrade I can still change them^^
    Does anybody tried the Infineon BD139, BD140 ? 
    rayshader likes this.
  10. ringer
    I want to change 4700uF caps (that has dimansions 30x30 mm) on bigger caps 30x40 mm. Is it fit in Lovely Cube body?
    (Sory for my English. I have a little talking experience.)
  11. francisdemarte

    BlaBlaBla has a great point. If amp works properly and sounds great why change anything? Just enjoy it.
    Those of us that insist on using genuine parts, name brand parts, Stephen sells the LC as an unpopulated board. You can hand select all the other parts from a trust worthy vendor and build it according to the parts list (or overbuild it) yourself. This is what makes the LC more attractive than the M-Stage for the technically inclined.
  12. BlaBlaBla
    Ringer, yes they will. For example Mundorf Mlytic SI/PI are 30x40mm. They fit in. After trying both of them, Mundorf eat Nichicon for starters.
    I was reading what Fliukell did, and because I really like to do experiments, and I can not help myself not to do that. 10 minutes ago I replaced Mundorf SI with stock Nichicon and right now I ask myself two things... First, where the low register has gone, second why I replaced good part with this crap (so the entire low register has gone). There is no punch, there is no grunt, everything is so mellow. Lower frequencies are so shy. When there is bass, it has less control. I mean, if you like mellow bass and huge sweetness - it is ok. But this is not for someone who like "stage in front of his face"...
    Probably that is reason why Mundorf cost 15 USD each, and Nichicons 6 USD each.
    Second tought. Do not use WIMA for coupling on output. You will think that someone is speaking from the bottom of his butt. They emphasize depth too much, and voices are artificial and metallic. With sibilance. Only if music is perfectly recorded, this setup will have benefits because richer space. Stock ERO is "in front", slightly flat, sometimes rough, warm, and has big separation from left to right (wide)... Good for old recordings. Even if it is not genuine...
    Now I have to re-solder those parts back.
  13. BlaBlaBla
    Re-soldered. Wima is out - ERO KP is back. Separation is, again, flawless.
    Probably, for three or four weeks it will arrive another Lovely Cube PCB.
    Also, I ordered PRP (signal), Takman (1W) and Kiwame (2W output) resistors (among with LT317/337 regulators) from Canada. When those parts arrive I will solder them on new LC PCB, and then I will be able to compare those two PCB-s (one with 50% stock parts - that I have right now, and LC that will only have stock Alps pot)...
  14. ringer
    Thank you. Already ordered Mundorf M-Lytic PI 63 VDC 10000 uF.
  15. BlaBlaBla
    Not 10000uF 63V!
    6800uF 63V is maximum. 10000uF 40V also fits.
    But 10000uF 63V does NOT!
    I did not tell you exact dimensions that fit in a case.... 30x40mm is 4700uF 63V, 10000uF 63V is 35x50mm....

    Could you recall it, or change it for less? 4700uF is enough even for the most demanding headphones. Usually people say that 1A goes for every 2000uF max. 4700uF is enough for  2.2A on each channel. If you load output with 2A on each channel, you will probably burn LM317/337 before Mlytic falls down. They have sick discharging characteristic.

Share This Page