Device selection. Further small gains may be obtained by careful selection of higher gain versions of the BC547 and BC557. More technical members may wish to investigate the use of the B and C versions of the BC547 and BC557, or appropiate members of the same family of devices, including the BC548,BC549,BC558,BC559, BC550B,C and BC560B,C. It is best to use devices with gain (HFE) not much more than 600, and try to use different polarity devices with reasonably close HFE in each channel,as well as the other channel. i.e. The 2 x BC547 (etc.) should if possible, have a similar gain,and also similar to the 2 x BC557 (etc.)
Capacitors. The Suntan capacitors supplied with the Jaycar kits appear to be of quite good quality, and there is little to gain by using brand name "Audio" type electrolytics. In fact, the use of them may even accentuate the higher treble, leading to added sibilance. The bipolar input capacitors should be replaced by higher quality film type capacitors. The Wima 4.7uF MKS2-XL capacitors are a good choice , as they have 5mm lead spacing. Polypropylene capacitors will give slightly better results again, but are physically awkward to implement because of their bulk. (See Reply#8)
NOTES. Silicon Chip Headphone Amplifier Kits - Notes for A.K. version. You only need to turn the heatsinks around when using the 2SA1930 and 2SC5171 instead of the original BD139 and BD140. When the better specced Toshiba devices are fitted , the 100nF capacitors should be relocated under the PCB at the same location. RE the 4R7 resistors, 4R7 .25W are O.K. there too. However, if you are using the original circuit, you can just use what is supplied with the kit. I felt that the 4R7 1W that was supplied instead of the originally specified 4R7 .5W, was reducing protection to the circuit in the unlikely event of a fault condition.The larger sized 4.7 ohm resistors will prevent the heatsinks from being rotated. The fuses are bridged to further reduce power supply impedance. The basic changes to the Jaycar HA kit for my version are : (1) 4 x 4.7ohm 1/4Watt or 1/2 watt resistors to replace the larger 1W type supplied. (2) 1 x LM4562 DIP8 (not SMD version) (3) 2 x 2SA1930, 2 x 2SC5171 (Toshiba) Mike's version uses 2 x 10uF film type capacitors to replace the 10uF Non Polarised electrolytic capacitors supplied. Mike will be best able to advise on other component choices for his version, including additional mains conditioning. Don't forget that you will also need other items such as a suitably sized case, 15-0-15VAC 20 or 30VA transformer, a male IEC fused type mains socket, 1 x M205 .5A fuse, on/off switch 1 "red" RCA socket, 1 "black" RCA socket, 1 x 50Kohm dual logarithmic potentiometer, and a suitable knob for the potentiometer. For better results, an Alps 50K Blue Velvet type is recommended. It not only sounds better, it also gives better tracking between channels at different volume levels. A 50Kohm dual attenuator will further increase resolution if used to replace the Alps potentiometer. e.g. www.diyfidelity.com.au/product_info.php?cPath=0_33&products_id=32&osCsid=efccc5aec9b830994a420a6e10d7148c
The voltage regulators provide current limiting at something over 1.5 Amps,as well as having internal thermal limiting. The originally specified .5W resistors would have given further protection, but it appears that low value .5W resistors aren't normally stocked these days, so Altronics and Jaycar substitute 1W resistors. As the modules are unlikely to draw more than a couple of a hundred mA on peaks, even .25W resistors appear to be suitable. In fact, I have been using the normally hard to get 4.7 ohms .25W metal film types. The Toshiba output transistors have a typical HFE between 150-200 at 100mA, and a transition frequency of 200MHZ typically. The BD139 and BD140 have a gain as low as 40 at 100mA, and depending on the maker, an Ft as low as 50MHZ. The higher gain Toshiba devices also reduce the loading on the previous stage, which is beneficial. They should be fitted with heatsinks, although they are a little more rugged than the BD139 and BD140. They are also reported to have excellent linearity. Also , the NPN and PNP devices are usually closer in HFE than the BD139 and BD140, which again improves linearity of the output stage.
P.S. I have found that the Cat.no. ZD0120 3mm green diffused LEDs from Jaycar, have a forward voltage of approx. 1.85V at 1mA, and are a suitable replacement for the RED LEDs supplied in the kits, if anybody just wants to be different.
JLH SCHEMATIC The recommended 2,200uF 10V low ESR capacitors are Suntan CD286 .Jaycar cat. no. RE6306. Several U.K. members are now also using these. The actual PCB layout will depend on the supplied version of the PCB. The 2.2uF capacitors may be replaced by 1.5uF, or even 1uF, if availability is a problem.
JLH PSU Add On - Components listing.
1 x mini JLH PCB
4 x 1.5uF (or 1uF) 63V. with 5mm spacing between leads.
2 x 10uF 25V (or higher) electrolytic capacitors.
2 x 100uF 25V (or higher) electrolytic capacitors
4 x 2,200uF 10V (or 16V ) Low ESR capacitors. Check that they suit PCB layout.
2 x 33ohm Metal film .5W resistors
2 x 1K .5W MF resistors
2 x 2.2K .5W MF resistors
4 x 2.7K .5W MF resistors
2 x 10K .5W MF resistors
8 x100K .5W MF resistors
2 x 120K .5W MF resistors
2 x BC550C transistors.
2 x BC560C transistors.
1 x BC639 transistor.
1 x BC640 transistor
2 x 2SA1930 Toshiba transistors.
2 x 2SC5171 Toshiba transistors.
NOTE. BC547C, BC548C and BC549C may be used if BC550C isn't readily available.
BC557C ,BC558C and BC559C may be used if BC560C isn't readily available.
Jaycar CAT. NO. HM3173 RRP $1.55 (or similar) 3 WAY PCB MOUNT SCREW TERMINAL - 5MM PITCH
PSU LAYOUT with mods.
D1,2,3,4 are replaced by either UF4002 or UF4004. (I use the even faster SF12 diodes.) A 470pF 100V (minimum voltage ) film capacitor is connected across the 2 outer terminals of the 3 pin AC Input block. This appears to further reduce diode switching "noise". The recommended PSU is avaiable from Jaycar as a kit for AU$19.95 Cat no. KC5418 www.jaycar.com.au/productView.asp?ID=KC5418&CATID=25&form=CAT&SUBCATID=557 e Volume Control Wiring. The attached diagram is from the original magazine article, but will also have been included with the purchased kit. If the onboard RCA sockets are fitted, as in the original article, RCA plugs can be connected to the cable coming from the earth and centre tag of the volume control as shown in the diagram. If the onboard RCA sockets are not used , then the cable may be terminated under the PCB where the RCA sockets would have been. The shield of the cable is connected to the outer position of the socket location. Alternatively, if you have a PCB drill, a couple of PCB pins can be fitted to the PCB and the cable connected to them. Alex
Ok Pink Floyd (where do I know that name from,..?) got the ball rolling in the UK with some prodigious OP-AMP rolling hinted by SandyK also the PSU is pretty much done back on page 1.
Post 52 sees some minor tweaks moot point for me but removal of zobel network and a variation of the feedback pole frequencies summed up by SandyK page4 post #52 also introduced here is the JLH well it's a noise reduction device to fit between PSU and Amplifier module, read on,..... post #58 ish
Leo spots similarities to a commercial rig, page 10 post #139
Some talk of Canjunkie setting up a chemical lab, ... no to make PCB's for JLH noise reduction device. page 11 post #151
Pink Floyd gets itchy Op Amp fingers (again!) #163 LM4562's out AD843 fried and AD8066 in the electric chair ouch!
Some people have stamp collections Pink Floyd just shows off,.... page 13 post #187
Holland make some interesting revelations regarding indiscriminate opamp rolling input offset and other gotchas page 15 post #212
Pink Floyd discovers the joy and benefits of local bypassing at the OPAMP page 16 post #228 yes it does work well almost Leo set him straight
Offsets anyone? page 16 post #235
Canjunkie discovers PCB etching and a new JLH is born
MERTON discovers the thread, down hill from here folks post # 261
vicious rumors about the SC HA not being available anymore UK source of PCB's added by Insomniac page 20 post #289
Ok Alex that's a hard act t o follow, for the adventurer thrill seeker out there. I went a different path. I still want to try one of those JLH's haven't yet but the PSU I'm currently using is already pretty damn clean
So I concentrated on applying standard practices to local decoupling and supply filtering and tidying up stray coupling, the unit surprisingly still works actually it is more lucid sounding and seems cleaner then the standard SC HA, which I also have as a reference.
The topology is straight forward and not uncommon in fact similar to an Analogue Devices tech note design and many others, this one features constant current source bias on complimentary emitter followers. Overall feedback with a voltage gain of 2 in standard configuration which will restrict certain OPAMP rolling namely to those specified stable with a gain of one!, some OPAMPS are not stable at this gain setting! Incidentally for an LM4562 the little fellow dissipates about .33W (~11Ma * 30V supply!) and that's sitting there at idle folks.
Personally I feel the input design and layout is somewhat ordinary, no it's krap! way too much opportunity for stray coupling I'm sorry if that offends but seriously you might as well call it an antenna farm I could easily get the original to oscillate using finger coupling from transistor heatsinks to the body of the OPAMP, not quite a theremin but!
Doubly so with the heatsinks on the transistors acting as capacitor plates is only asking for HF trouble with wide bandwidth OPAMP's and transistors. So from a mechanical point of view, I don't like seeing transistors flapping in the breeze with heatsinks attached so a bracket was made to secure the devices to, as well this provided a common point ie connected to ground (0V). So that's why I bothered doing it, now the unit is un-conditionally stable, keep in mind there is only one pole in the feedback arrangement and that only reduces the gain to around 1 at higher frequencies.
Bias diodes well what better then to use a transistor junction to track temperature, a transistor of the same type! nope that wasn't going to happen, even better have the diode on the same die! A BC547 is what we got, Ok the voltage drop across the PN junction's is a little higher then the 1N4148's so standby current goes up likely not too bad a problem and the (diode) transistors are attached to the bracket so thermally track at ~40mA each. The constant current source has had 6V2 zeners and the decoupling capacitor increased to 470uF to further reduce coupling from supply to current source i.e. improve low frequency PSRR
Supply is fairly simple 100nF bypass at input and 220uF 1mH (it's actually a swinger 4mH to 1mH 1mH@1A) 220uF Pi arrangement which with the original supply decoupling taken to 1000uF makes for a squeaky clean and low impedance supply with the idiot diodes just for my peace of mind as the wiring goes to 4mm banana plugs which plug into the "Lab PSU" Further supply decoupling has been added to the OPAMP by dropping resistors and zener diodes with tantalum and mkt capacitors to improve an already impressive 110db PSRR of the LM4562 it also means that for all practical purposes the OPAMP can never remove bias from the output transistors so they always remain on
With the output transistors local decoupling is added directly form collector to ground by supplementing the 100nF with Tantalum 4.7uF to ground at the collector(s) under the PCB, that solder mask is damn hard to remove this further improves HF stability!
Which leads me to the inclusion of the Zobel networks for me it's just giving some isolation from the many and varied reactive loads that could be plugged in the the amplifier, so I kept it besides doing the maths there is no indication of effects in the audio band and beyond. This is likely a personal preference, as a telling feature I discovered that there is some residual 1.6Mhz from computer o/p only a couple of mV pp verses a few volts of signal, well the amp has a gain of about 1 at that frequency and the residual 1.6Mhz is on the emitters but not even measurable on the output jack!
I'll have to fess up with the current limit I have none at the moment as the PSU takes care of that but eventually the amplifier will have to stand on it's own feet with it's own PSU so this is a work in progress but follows fairly standard design practices, so stay tuned
As a passing comment don't take what the silicon chip article says as the theory of operation to seriously as reading it some of it makes sense then the rest of seems like they got it off the back of a Cornflakes packet,.... so I expect the design is something they modified from somewhere else, nasty I know but hey
Input Attenuator. Several members are now using Attenuators at the input instead of a Volume Control. Users report that there is a little bit more extra detail, and it is akin to a small veil being lifted. The downside, is that there are only 23 or 24 positions available with the majority of attenuators , however most people who use them are happy to live with this minor consideration due to the improved performance. Attenuators vary greatly in price, and to a lesser extent, performance. A few people (including myself) are using a kit from DIY Fidelity, which can also be purchased ready made for an additional AU$25. Alex
Nick Unless you are really determined to put the attenuator together yourself, I suggest you pay an extra $25 dollars, and let them make it for you. DIY Fidelity is based in Sydney, and I ordered mine on the Thursday before a long weekend. They made sure that I got it in on Friday, even though I didn't ask for that. I have soldered 10s of 1,000s of soldered joints during my career, but assembling this is damn tedious, and easy to make a mistake. BTW, it uses Vishay/Dale Military Grade 1%. I hope you have a magnifying glass and a good Digital meter. They aren't colour coded, and the values are unusual compared to what we are used to. Alex
Replacing the Input capacitors with 4.7uF Polypropylene capacitors.
It's quite simply done if there no RCA sockets fitted . No holes need to be drilled. You stand the polyprop vertically with the bottom lead going into the outermost mounting hole for the RCA socket, (in line with the original capacitor mounting holes) The leads at the top are extended slightly by using the leads from a spare resistor, sleeving the lead with a piece of 3mm heatshrink, (or similar after soldering ) then pushed through the inner most hole of the original capacitors location and soldered. The input leads are then terminated to the underside of the PCB.Actually, it doesn't look too bad at all, and of course sounds better than the original bipolar electro or a 4.7uF Wima. I used 4.7uF 250V capacitors from Jaycar. Cat. no. RY6954 Alex
Click on the image for a larger photo,then click on the new image for a full size photo.
Yesterday at 10:05pm, ngquan wrote:Hi Alex, how much better with those 4.7uF polyprops in place compare to stock NP. I see the catalog from Jaycar-they are 8 bucks each!!!.
Leo wrote : Sorry, I'm not Alex but a film cap should sound better than any electrolytic in this particular application where the cap has such small DC bias across it. Of course all film caps are not equal so some are going to be better or worse than others.
I concur with Leo. Initially I replaced the bipolar electrolytic capacitors with 4.7uF Wima MKS2 film types,and noticed a definite improvement in all areas. After Leo remarked to me recently that the Wima 10uF in his new discrete Class A preamp sounded like shit compared to no capacitor, I decided to try these polypropylene caps from Jaycar instead. There is less colouration, and the bottom end sounds a little better, as well there is a small reduction of sibilance in some recordings, but no apparent loss in resolution. In my unit with the attenuator, the soundstage also seems a little better, with a small improvement in localisation of instruments. Results will of course vary with the type of headphones in use etc. Alex
Recommended Mains Wiring. SAFETY WARNING. The strip on these I.E.C. fused sockets is ALIVE at ALL times. Please either fit an I.E.C. socket shroud over the socket, or cover the strip with insulation. I covered mine with a layer of Araldite before installation. The fuse used should be a 0.5A fast blow, M205 type. An even better method is to use a non fused I.E.C. socket, and install a round M205 holder with a 0.5A M205 fuse. e.g. Jaycar CAT. NO. SZ2034 M205 Safety Fuse Holder - Panel hole 12mm. RRP $1.40
The Active (A) of the I.E.C. socket would then connect to the fuse holder, and the transformer primary winding wire (usually, but not always brown) is connected to the other side of the fuse holder,either directly, or via the toggle switch if a DPDT toggle switch is used. All mains terminations should be covered with heatshrink tubing. In the case of the fuseholder, it would be a good idea to cover the body with a piece of slipped on, then shrunk, larger diameter heatshrink tubing as well. A pack of assorted heatshrink tubing is always very handy to have. e.g. Jaycar CAT. NO. WH5520 The Ultimate Heatshrink Pack This must be the ultimate heatshrink package. - 1 length each of 7 different colours in 7 different sizes ranging from 1.5mm dia to 20mm. - You may never need to buy heatshrink tubing again!- Sizes: 1.5, 3, 5, 6, 10, 16 & 20mm. QTY 1+ $11.50
Mains Earth Terminal. (reproduced from Silicon Chip magazine,in the interest of the safety of constructors of their designed kits)
Click on the image for a larger image, then click on the new image for a full size image.
If you wish to use 8 ohm headphones with this amplifier, it would be best to use the original design or RobertKD's version. Output level MAY be be too low with the JLH version, although a JLH could be used with the original version without any circuit changes. PLEASE NOTE. The original design should be modified as per the published article, if 8 ohm headphones will be used. This is mainly to help prevent hearing damage. I presented the simpler version that could be used with the vast majority of headphones without any circuit modifications. Alex
P.S. If there are any aspects of this version that members feel hasn't been covered sufficiently in this "Tweaks" thread, please ask in the original "Jaycar" thread, or contact me. If you have any suggestions for further improvements, additions, or modifications to this version, please send me a PM or an email.
Any queries regarding Mike's version, or Robert's version should be directed to them.
With the SC HA version using the JLH, it is possible to have independent outputs, of different impedances if required. One way to achieve this ,could be to utilise the area vacated by the no longer used R and C of the removed Zobel network. PCB pins could be fitted at those locations on both sides of the PCB, and a couple of short lengths of shielded cable used to connect to an offboard located headphone jack. The shields would be connected together at the aux. headphone jack. The 68 ohms resistors,(or selected values) would replace the original 10 ohms resistor of the Zobel network,either directly, or via added PCB pins at those locations,making it easier to alter their resistance value. The existing 68 ohm output resistors would be retained for the onboard headphone jacks(s). The only drawback seems to be that you need a PCB drill to enlarge the holes slightly to take a 1mm PCB pin. Alex
Robert I agree. That is why I suggested only very modest changes. Obviously, by reducing the 4K7s feeding the LEDs to 4K3 (or3K9), there would only be a minor increase in LED forward voltage. Likewise by reducing the emitter resistors from 100R to 91R, you would only get a 10% increase.Not a big deal going from ,say 30mA to 33mA ? Alex
A small improvement in resolution and overall SQ, may be obtained after the extra resistor is added. The amount of improvement will depend on how well matched the Differential Pair transistors are. The modification may be implemented by carefully cutting the appropiate track with a sharp instrument such as a scriber. The additional 2.7K 1% MF resistor, is then soldered between the supply rail and the collector of the transistor involved. (the other transistor already has a resistor) Carefully check your work afterwards to ensure that there are no solder bridges to nearby joints. The modification is performed underneath the PCB on both sides. Alex
what about the difference in voltage dividers even with 1% R's it could still be +/- 150mV? so I thought maybe
Robert I agree with what you are saying, but if the devices are reasonably close in HFE, the addition of the resistor should still give better results technically,if not audibly. I buy my MF resistors in packs of 8 from Jaycar, and measure them before installing them. Most of the time, the majority in the pack are very close together in value. Most people wouldn't go to the trouble I go to with matching,and that's why this suggestion was pitched at people who have made some attempt to match pairs. It's going that little further at each step that makes the end result better. Actually, many supplied devices will be from the same batch, so it would not be normal to see too large a divergence in HFE and VBE, so many constructors may already have pairs that are relatively similar. Many constructors will also have a DMM with transistor test facility. The JLH that I just updated,actually has approx. 50mV difference between collectors on both rails, but although the LTP devices were originally selected to be reasonably close in HFE, they were not specifically matched. Don't forget though, that we aren't talking about a signal amplifier for listening, we are using this to give a clean amplification of noise/ripple to eventually be used for cancellation purposes.I find the fact that it can actually influence SQ so positively, quite intriguing, as we are doing a helluva lot more with this PSU area than most designs attempt. We obviously need to consider the PSU area as a major part of the signal chain. Another reason for modifying my own unit, was that Miguel's replacement, using the trial changes, sounded a little better than mine ! My original JLH PCBs had provision for this change, but I never got around to trying out my idea previously, and just fitted a link.
Alex P.S. We could go to the lengths that you have shown, but I doubt that there would be much further audible gain,and a PCB redesign would be required.
I have been doing the same with the emitter resistors, just matching them as close as possible (given these a only 5% tolerance), also adding R in series with the bias diode to change the Vbe my aim is to have the output stage as closely matched as possible so there is zip offset before adding the opamp to servo the output. Will it make any difference not sure but hey figure it's worth a shot.
Robert Atttention to detail usually pays dividends. Not always large, but usually worthwhile. Not every apparently worthwhile idea works though. For example, in the last couple of days, Leo and myself swapped the 10uF coupling electro in the SC HA's JLH with a 10uF Wima. After a couple of days it stabilised , but I found it sounded a little "plasticky" in midrange. We both came to a similar conclusion independently, and have both now reinstalled the original 10uF electro. It wasn't an earth shattering difference, but the electro had less apparent colouration and better dynamics in that part of the spectrum. Leo also found that a 10uF Wima in the input of his build of the SC Class A preamp sounded like shit, compared to no input capacitor, however it did sound better than a bipolar electro. Alex
have been doing the same with the emitter resistors, just matching them as close as possible (given these a only 5% tolerance), also adding R in series with the bias diode to change the Vbe my aim is to have the output stage as closely matched as possible so there is zip offset before adding the opamp to servo the output. Will it make any difference not sure but hey figure it's worth a shot.
I do likewise with emitter resistors ,(similar in both channels too!)but not quite as fanatical with the biasing diodes. I have a pack of 1N4148 on a roll,(economical AND handy.) and quite often use matching diodes with a slightly higher than average forward voltage drop.
Chris Yes. If you look very closely at the photo I posted of Miguel's unit (M1Am1), you will see 2 tiny MF resistors at that location, against the 2.2uF. It is easier to shove them across the cut track under the PCB than drill holes like I did that time. Alex