I added the output latchup protection diodes as described in Post #2668. I believe the rest of the circuit topology is identical, unless you see something I have missed.
Resistor values shown in my published schematic were altered to increase parts commonality (fewer BOM line items) or to avoid values not stocked at popular distributors. As mentioned at several points in the thread (e.g., Post #2593 and Post #2432), component selection and values are not critical. As I recall, Scott mentioned that he assembled his concept-validation circuits using parts on-hand rather than the exact engineered values, specifically to verify that component types and values are not critical.
I didn't include the thermal coupling shown in Post #2591. Tight thermal coupling is very difficult, to impossible, with discrete SMT parts. As mentioned in Post #2598 the lack of thermal coupling isn't expected to be a serious disadvantage.
THERE IS A DRAFTING ERROR IN THE POST #2643 SCHEMATIC!
The labels for " +in " and " -in " are transposed. You can verify this by (carefully) comparing the topology of Post #2643 to the schematics attached to Post #2409 and Post #1568. This is also apparent when you simulate the Post #2643 schematic.
My published schematic includes an input stage balance adjustment (R28 and associated components). This adjustment was not shown on Scott's schematics but the function was discussed in posts (e.g., Post #2667 and #2668). I selected component values that seemed reasonable.
My published schematic includes the secondary compensation components (C2/R10, and C3/R14). These are shown in the Post #1568 schematic but removed from later versions. I retained the networks to provide the greatest flexibility for customizing the compensation. The locations may be left unpopulated if a simpler compensation strategy is used. The values I show give reasonable simulation results with the models I have for the active devices shown.
Similarly, I made provision for moving the secondary compensation networks between the two sides of the differential input stage. I discussed this in Post #2404 et seq.
My published schematic implements the compensation networks as series RC (i.e., pole-zero) networks. This is consistent with the Post #1568 schematic although only a capacitor is shown in later versions. I retained the RC network to provide the greatest flexibility for customizing the compensation. To implement a capacitor-only compensation scheme, install low-ohm resistors (e.g., 1R0), or zero-ohm jumpers, or found-on-bench (FOB) scraps of hookup wire in place of the resistors.
My schematic uses diode-strapped transistors as the level-shift devices (Q7, Q8, Q9, Q13, Q14, Q15) in the input stage. This is in agreement with Post #1568, but was altered to pure diodes in later versions. No claims of improved performance were made for diodes versus diode-strapped transistors. In practice, the component footprints on the PWB will accept diodes in SOT-23 packages at those locations - e.g., BAL99 or BAL74 depending on the polarity.
The Post #2643 schematic shows an LSK389 matched-pair JFET as the input device. Other versions used 2SK170's or BF862's, and reported acceptable performance. I retained the SOT-23 single devices because the LSK389 is not widely available from hobby suppliers. Also, a PWB designed to use the BF862 in SOT-23 package is compatible with other JFET models in SOT-23, where the LSK389 footprint offers very few (if any) alternatives.
Throughout my implementation I used single transistors in SOT-23 packages. (The output devices may, optionally, be SOT-223 devices.) While using dual devices may improve thermal tracking and save PWB real estate there is a much wider selection of suitable parts available in SOT-23 than in comparable duall-device packages. Again, this approach caters to flexibility in construction on a common PWB.
Considering the above, I believe my implementation is compatible with the Post #2643 and Post #2668 schematics. If you have identified other differences please provide specific details. It is possible I have overlooked or misunderstood some aspect of this circuit.
Dale
I dont believe anyone is forcing you to use SMD, but there is performance improvements in less inductive, physically smaller parts and layout, that would be why they made the first ICs to begin with wouldnt it?
hmm thats why I asked Scott, it will certainly have impact objectively; I just wondered how much in this specific case
I wouldnt expect you to do anything else
The lead length from the resistor body to the pcb is pretty short..... that would be the only difference as the pcb trace length and shape will dominate. A smd without leads can be shorter than a leadless one but the pcb traces dominate.
I am not as convinced for audio freq range that if you measured thd, s/n, Im, freq resp et al that they would be different due to sm vs thru-hole. I'm willing to be proven wrong.
Thx-RNMarsh
I am not as convinced for audio freq range that if you measured thd, s/n, Im, freq resp et al that they would be different due to sm vs thru-hole. I'm willing to be proven wrong.
Thx-RNMarsh
Last edited:
It just seems to me that SMD is just not for the DIY assembler where he is only making maybe four units for a stereo preamp. It is for the manufacturer that is making hundreds or thousands. What DIYer wants to invest in a 40x stereo assembly microscope and a reflo oven to make 4 units? Also, Scott said the unit wouldn't support being a unity gain follower with the + input grounded. The JE990 will support being a unity gain follower. Ray
none of the parts being used is that small, none of the parts used needs reflow, all are quite solderable with a standard iron and a standard size tip. 40x stereo assembly microscope for this? try a standard 4x $10 illuminated visor if anything at all. SMD is MORE friendly to the DIYer once you get used to it, PCBs are much easier to make if you dont have to punch a heap of holes in them, especially something this size.
fine if you dont want to do SMD, but your reasons as posted do not make a lot of sense IMO. 1x or 1000x its just as suitable.
fine if you dont want to do SMD, but your reasons as posted do not make a lot of sense IMO. 1x or 1000x its just as suitable.
I don't think you need a microscope for that kind of job.
Better get used to SMD, soon no through hole components will be available on the market.
soon no through hole components will be available on the market.
I can solder 0805 & larger without an aid, but use a jewelers loup (10X) for anything smaller.
SMT is advantageous for a smaller PCB area = lower cost & automated assembly but you are usually forced to use a double-sided boards minimally and almost always need vias. Almost everything I design is mixed technology to some degree or another, especially if it is analog, for digital, well that is usually denser, thus SMT, is the primary component choice for so many reasons.
Keep up the good work folks!!
Rick
You can say this about certain newer small signal semiconductors & IC's, but probably not for power semi's nor many passives.
I can solder 0805 & larger without an aid, but use a jewelers loup (10X) for anything smaller.
SMT is advantageous for a smaller PCB area = lower cost & automated assembly but you are usually forced to use a double-sided boards minimally and almost always need vias. Almost everything I design is mixed technology to some degree or another, especially if it is analog, for digital, well that is usually denser, thus SMT, is the primary component choice for so many reasons.
Keep up the good work folks!!
Rick
Last edited:
I always work small parts under a stereomicroscope, and I can tell you, 40x is way too much. 4x magnification is more what you'd want. Look for second hand buys of a good brand, and it won't cost you a fortune. One of the best investments in workshop tools I made.
I once had the chore to make another designers circuitry work - who quit just before recieving his design for testing -- It took me a full year of daily trouble-shooting and interface testing to computer data acq and control and the like for an In-Situ coal gasification proof of concept experiment. It used 300 digital IC's. All wire-wrapped to one another. 
A one-off design.
I can appreciate the use of smd and high integration of circuits! But, I am not too worried about the future as it is going more and more digital... with just input interface and output (maybe) being analog. And, input can be IC. So, I wont have to worry about building descrete analog audio any more. And, no worry about home made one-off DIY products with descrete smd, either.
Enjoy the music.
-Thx RNMarsh
A one-off design. I can appreciate the use of smd and high integration of circuits! But, I am not too worried about the future as it is going more and more digital... with just input interface and output (maybe) being analog. And, input can be IC. So, I wont have to worry about building descrete analog audio any more. And, no worry about home made one-off DIY products with descrete smd, either.
Enjoy the music.
-Thx RNMarsh
what are you telling me for?
My questions were rhetorical; I didnt mention the need for 40x, that was dchisholm. I have a 4x illuminated visor, but I hardly use it since for sot323 etc its not needed and for larger QFP types I use drag soldering, so seeing each pin is not of any importance. I was saying 40x was too much and that 4x was about right; if anything is needed. I mostly only use it for checking quickly, but I find taking a high quality macro photo and viewing it in photoshop is more useful for error checking anyway.
40x visor would be impossible to use, move the piece or the iron slightly and it would go from one side of the viewer to the other.
Last edited:
A good alternative for a stereo microscope could be an USB microscope, see:
Dino-Lite-- Digital Microscope
Essentially a high magnification real time video camera. it takes just minutes to get used to work watching the PC screen instead of the real circuit board directly. I've learned this from our ECM that routinely does reworking jobs on very hi-tech boards.
Regards
Giorgio
Dino-Lite-- Digital Microscope
Essentially a high magnification real time video camera. it takes just minutes to get used to work watching the PC screen instead of the real circuit board directly. I've learned this from our ECM that routinely does reworking jobs on very hi-tech boards.
Regards
Giorgio
of course, but my point was its not a requirement as suggested, in fact 40x is not even useful for hand soldering and not so cheap, that would make the smallest PCB under discussion here appear nearly 1m x 1m... a single sot23 100mm x 100mm
Last edited:
Intriguing suggestion. I have recently used a visor magnifier but it's pretty disruptive despite freeing both hands in principle at least.
S.A.G.,
My only question is how is one of these cameras different from just an ordinary video camera that we have on a computer? I have a fairly good camera for video on my computer, a top model Microsoft camera. Why not just use the software on an existing camera with what looks like an led light to light up the board you are working on. I assume it would take a little time to get use to looking at a screen and manipulating and iron at the same time.
My only question is how is one of these cameras different from just an ordinary video camera that we have on a computer? I have a fairly good camera for video on my computer, a top model Microsoft camera. Why not just use the software on an existing camera with what looks like an led light to light up the board you are working on. I assume it would take a little time to get use to looking at a screen and manipulating and iron at the same time.
qusp,
I have never tried to use the camera for anything close up so you are probably correct. I know it has auto focus but not sure of the limitations of the lens or focal length since none of that kind of information is given. Sure would be nice to have an auto focus though and 4x power isn't much, I have better than that on my stereo visor now.
I have never tried to use the camera for anything close up so you are probably correct. I know it has auto focus but not sure of the limitations of the lens or focal length since none of that kind of information is given. Sure would be nice to have an auto focus though and 4x power isn't much, I have better than that on my stereo visor now.