First Watt Amplifier & PSU Build

Discussion in 'Audio Hardware' started by fully_articulated, Jul 19, 2020.

  1. fully_articulated

    fully_articulated Forum Resident Thread Starter

    For any members interested in this kind of thing, here is a thread on the building - or more accurately, rebuilding - of a First Watt clone amplifier. As I'm currently having to spend a lot of time waiting on parts (which due to geography can take up to 4 weeks to arrive) I figured this might be a nice way to fill in the downtime.

    For those unfamiliar, First Watt is kind of a sister company to Pass Labs, both selling the products of owner and designer Nelson Pass. With First Watt, Mr. Pass indulges himself designing low-power, simple circuit, Class A amplifiers. He will also, very generously, release the schematics of these amps for use by the DIY community.

    About two years ago I constructed my current F6 amp thanks to the guide and enormous threads found on the diyaudio forum. Seen below, it is a relatively standard example of its type, utilising the boards sold in the diyaudio store along with the recommended case and transformer.

    [​IMG]

    DIY can be fun, especially since you get the benefit of listening to (and hopefully enjoying) your efforts once done, but I also try to treat each project as an opportunity to expand my limited electronics knowledge and attempt things I haven't done before. The main focus of this rebuild is the power supply. More an alteration than a full redesign it incorporates a few additions to the standard supply, a change in filtering type, and some custom parts.

    Here is the project in its current state.

    [​IMG]

    The circuit board wrapping around the transformer is my first experiment in PCB design, and it seems to work, but hasn't been tested under load yet. For the filter section, two large inductors sit at the rear of the case and the empty area in the centre is the future location of the capacitor bank which is waiting for the delivery of a custom mounting plate.

    This is the planned layout for the amplifier.

    [​IMG]

    I'm not totally sure how this thread will proceed. There's certainly more build photos to be posted. It may also be worth sharing some of what I've learned about PSUs in general, if that's to anyone's liking.

    Lastly, and rather self-servingly, my hope is that other forum DIYers will let me know where any of my info/assumptions are wrong, if I'm making any construction blunders or poor choices etc. Any feedback will be greatly appreciated.
     
  2. tumpux

    tumpux Well-Known Member

    Location:
    doha
    Neat, if I may know how much is the total cost of that amplifier so far?
     
  3. fully_articulated

    fully_articulated Forum Resident Thread Starter

    Nearly all the parts have been purchased now so I can give a rough idea of total cost, but bear in mind it's significantly more than one would need to spend.

    The chassis, custom faceplate, and internal metal parts (some also custom) came to $1350 NZD ($885 USD). The chassis on it's own is available from Modushop for $240 USD, so obviously it's the custom metal which elevates the price so much. Unfortunately I have neither the equipment or skill for my own metalwork.

    It's hard to pin down the component cost as easily, as I will usually buy more than needed as spares and for future projects. With this in mind, the PSU parts were around $1200 NZD ($785 USD).

    Plus there's hardware items like nuts & bolts, wire, connectors, etc that you might buy 50 of when you only need 12 for example. And then there's the things you order by mistake....

    I have read that the total cost for a standard build being about $700-800 USD so it's a bit embarrassing to put it all down in print like this, especially when it doesn't take into account shipping costs.
     
  4. fully_articulated

    fully_articulated Forum Resident Thread Starter

    The wraparound PCB was my answer to the problem of how to mount the additional components I wanted into the case. These being a fuse for each secondary winding, a snubber circuit for the transformer, and a way of attaching the rectifier blocks to a heat sink. And so, the Transformer Input/Output Board was born.

    [​IMG]

    It has seen many iterations and is perhaps the riskiest part of the new power supply. There are quite a few resources on the web for circuit board design but they were all very dense and technical. Much of that literature seems to focus on the relationships between power v signal, high frequency lines etc, which is a bit too complex for what I'm messing around with.

    Prodded in the right direction by a member on the diyaudio forum, my chief concerns became the safe clearance distance between traces and their required thickness. There are a number of calculators on the web for this purpose but I found the ones on these pages to be the most useful.

    PCB Trace Spacing Calculation for Voltage Levels
    PCB Trace Width Calculator and Equations

    Each page provides some good information without being overwhelmingly technical and the calculators use standards different than most online, usually producing more conservative values. I won't repeat the info found on those pages but the very basic rule-of-thumb is that the voltage dictates the space required between traces and current the thickness of.

    In that regard, I believe, the board is well over-specced. In fact, I was unprepared when beginning to populate the PCB for the amount of time and heat that was required, no doubt due to the relatively large amount of copper in the traces.

    [​IMG]

    Here is one 'wing' of the board with it's components soldered in. And below, the mains voltage section, which would usually have to be wired to a terminal block but could now be moved to the PCB also.

    [​IMG]

    The black thermistor used to limit the current spike at turn-on in the standard First Watt power supply is an Amphenol CL-60, but has been replaced here with an Ametherm SL22-5004 as it appears more suitable for the job. I also have an SL22-6003 that could be better still.

    With the TIO board complete (minus the rectifier heat sinks I was still waiting for) I was able to hook up the major components of the power supply for a bench test on the floor of my garage.

    [​IMG]
     
  5. fully_articulated

    fully_articulated Forum Resident Thread Starter

    Not finding an angle bracket or dedicated transformer bracket that suited my selected transformer, I decided to have a local metal shop produce one from my design.

    [​IMG]

    It worked out quite well.

    Making it from 4mm aluminium, they also suggested the addition of the gusset to increase strength. The only thing I didn't factor in was the bend radius reducing slightly the amount of flat space available on the vertical plate. This means the transformer only just sits flat against it.

    Afterwards I had it powder-coated.

    [​IMG]

    The Toroidy transformer is already a work of industrial art and the two pieces are quite beautiful together.

    Strangely enough, the bracket actually cost a bit more than the transformer itself if we don't include the cost of shipping from Toroidy in Poland. So, while nice, I'm not sure it's something I'd do again.
     
  6. fully_articulated

    fully_articulated Forum Resident Thread Starter

    Deconstructing the First Watt DIY power supply - Part 1

    I can't claim to know what the experience of putting together one of these amplifiers is like for others that have done it, but in my case I only had a very basic idea of what the function of the separate parts were and how they worked together when building the original. Regarding the amplifier sections, I am still almost completely clueless.

    In relation to the power supply, at the time I knew a 300VA transformer with dual 18 volt secondaries was required, but not why. I knew the output voltage needed to be 23ish volts - but again not why, or indeed how this was achieved. That the transformer took 240 volts from the wall and outputted 18 volts, that the rectifiers converted that 18V AC to DC, and that the capacitors & resistors formed a filter responsible for smoothing that DC somehow, with the end result being 23 volts, was the limit of my understanding.

    So, for all those would-be builders out there who may be in a similar place, I offer up my own basic examination of the First Watt PSU and its style of linear power supply in general. Hopefully the info will be accurate and to some, useful.

    First up, most of what will follow came from the below video series on power supply design. I just applied the teachings from them as much as I could to the spec of the First Watt design. What I particularly liked about this fellow's videos was how he talks you through the math and its application without it being too burdening.



    There is three more videos in this series.

    25 Watts of POWER!

    Many of the First Watt amplifiers (and all of the ones I'm interested in building) produce 25 watts into 8 ohms, although only one of them behaves in the way many might think of as 'ideal' by doubling its output into 4 ohms. That would be the F6. On the other hand the Aleph J power output decreases to 13 watts at 4 ohms.

    But, for the purposes of a universal PSU for any First Watt amp, let's assume the F6 as a base and also pretend that it is able to double its output again into 2 ohms, producing 100 watts. 25W into 8ohms, 50W into 4 ohms, 100W into 2 ohms. Disclaimer: I have no idea whether the F6 is capable of such a feat.

    By solving the square root of 25x8 we get 14.14 volts RMS (Vrms).

    This result is also true for 50x4 and 100x2, meaning the voltage at full power is the same no matter the speaker load. But, because audio signals are AC we must know the peak value of that voltage.

    [​IMG]

    Voltage peak (Vpk) is calculated by multiplying the RMS voltage by 1.414 (or dividing by 0.707 - whatever takes your fancy). The result for 14.14 volts RMS is 20 volts peak.

    If we add a few volts on for headroom and ripple, the 23 volts delivered from the standard power supply starts to make sense. Also the purpose of the bipolar nature of the PSU - i.e. having both positive & negative 23V rails - becomes clear as 20 volts AC is 20V above ground potential and 20V below, or 40 volts peak-to-peak.

    So, if our amplifier doubles its power output as the resistance load of the speaker halves, but with the voltage staying the same, what else is changing? Answer: current.

    Power divided by volts RMS gives us the current output.

    25 watts / 14.14 volts = 1.77 amps
    50 watts / 14.14 volts = 3.54 amps
    100 watts / 14.14 volts = 7.08 amps (multiply that by 1.414 and you get 10 amps peak)

    This is the second piece of the power supply puzzle. If we consider our conceptual F6 amplifer, one that can produce 100 watts into 2 ohms while running 23V DC rails, then the power supply should also be made capable of delivering 7 amps RMS and 10 amps peak when required.

    This is what I've learned. I encourage anyone whose knowledge extends beyond mine to fact-check any of what I've just written, and please correct me where I'm wrong.

    When my hands recover I'll try to explain how this info may translate to parts selection. Congratulations if you got through the whole post, and if it was of any help to you let me know.
     
  7. fully_articulated

    fully_articulated Forum Resident Thread Starter

    I'll try to keep the boring shots of chassis preparation to a minimum, but here's an image I quite like showing the solid aluminium feet attached to the bottom panel. I used these style feet in my last project and they grip very well. Available from many chinese sellers.

    The holes for the transformer bracket have also been drilled, with just a hand drill and some calm nerves.

    [​IMG]

    Unfortunately it all didn't go so well. A lapse in concentration produced the messy squiggle below.

    [​IMG]

    It got worse when attempting to drill holes through the grill slots.

    [​IMG]

    I had oriented all the internal parts so that their mounting holes would align with one of the ventilation slots, and hopefully then drilling into any awkward spots could be avoided. But even so, I found it difficult to keep the drill bit from chattering within the slot itself.

    Luckily these mistakes will largely be hidden once all the components are in.
     
    Rick Bartlett and Saint Johnny like this.
  8. james

    james Forum Resident

    Location:
    Annapolis
    Awesome thread. Thanks for sharing the build.
    Subscribed
     
  9. Agitater

    Agitater Forum Resident

    Location:
    Toronto
    Interesting project.
     
    Saint Johnny likes this.
  10. fully_articulated

    fully_articulated Forum Resident Thread Starter

    Thanks James. I appreciate the encouragement.
     
    Saint Johnny likes this.
  11. fully_articulated

    fully_articulated Forum Resident Thread Starter

    With luck it may even work :D
     
    Saint Johnny and Agitater like this.
  12. fully_articulated

    fully_articulated Forum Resident Thread Starter

    Fortunately the hole drilling was accurate enough to allow some hardware components a test fitting. First a trial placement of the circuit board.

    [​IMG]

    The heat sinks had now arrived and been temporarily mounted to the rectifier blocks. I have done absolutely no calculations on whether they are sufficient for their intended purpose - I simply found something that fit the space available on the board and will cross my fingers that it works when the power supply is delivering current. This particular heat sink is sold as a cooling solution for motherboard chipsets.

    My other concern is the rectifiers being mounted flat against the PCB rather than raised to allow air underneath. The board, rectifiers, and heat sinks will have to be monitored once the power supply is in operation for any overheating issues.

    Next the transformer was slid into place on its bracket.

    [​IMG]

    It was very satisfying to see the two pieces fit together so snugly. As hoped, they really do look like they were designed for each other.

    Unfortunately all was not well when viewed from the back...

    [​IMG]

    I have mispositioned the connector tabs of the high voltage section and one sits directly above the wires exiting the transformer, effectively making it unusable. I'll have to work out another way of connecting it to the primary wiring - maybe by soldering the cable in from the side.

    This, along with the less than stellar drilling of the chassis, made for quite a disappointing day of DIYing to be honest.
     
    timind, Manimal and Saint Johnny like this.
  13. Fdee

    Fdee Well-Known Member

    Location:
    Los Angeles, CA
    A right angled insulated Faston connector should allow you to use the blade right above the wire exits.

    I’ve built a couple of the first watt designs, using an M2 in my system now. It really is a wonderful gift that NP has given us!

    thanks for sharing your journey, will be following along.
     
  14. fully_articulated

    fully_articulated Forum Resident Thread Starter

    Ah, not even enough room for that I'm afraid.

    You're right, the DIY community certainly owes Nelson Pass a great debt. Thanks for your comments.

    Once I have this working with my F6, I'm planning on building the M2 myself along with the Aleph J. The plan is to use the power supply with those amps too, just swapping the amp boards in and out. I've designed it so desoldering anything should not be necessary.
     
    Saint Johnny and SandAndGlass like this.
  15. fully_articulated

    fully_articulated Forum Resident Thread Starter

    Deconstructing the First Watt DIY power supply - Part 2

    Transform yourself!


    To recap part 1: A theoretical First Watt F6 amplifier with a 23V power supply that, in a worst-case scenario, will be asked to drive 100 watts into 2 ohms, requiring current of 7 amps.

    To do this a transformer able to meet the challenge is needed. Firstly, to work out what the output voltage of the transformer should be, the specified DC voltage of the power supply should be divided by 1.414 to find the RMS value.

    23V / 1.414 = 16.27V

    Everything in the path from transformer to output will probably have some resistance causing voltage loss along the way. For example, the bridge rectifier I've chosen has a voltage drop of 1.1V by itself. Add that, and the total becomes 17.37 volts, close enough to make the recommended 18V seem ideal. In actuality, depending on the PSU makeup, transformer specs, and variations in line voltage the DC output from the power supply may end up anywhere between 22-25 volts in operation.

    More power to you!


    Transformer power ratings are listed in VA (volt-amperes), which perhaps unsurprisingly is calculated with volts multiplied by amps. Even though there is a difference between watts and VA, they are both measurements of power, and for our purposes let's consider them essentially the same.

    For a sufficiently detailed explanation: https://www.electronicdesign.com/ma...-the-difference-between-watts-and-voltamperes

    Anyway, by multiplying the desired transformer output voltage by the current requirement an approximate VA rating for the transformer can be found.

    18V x 7A = 126VA, which looks a bit on the low side doesn't it?

    Remember we have two 18V secondaries so our total will be double - 252VA. Giving ourselves some headroom means we might conclude selecting the 300VA value usually recommended for the First Watt amps is a solid choice.

    It may be worth noting that the equation can be reversed to find the current draw on the primary side, simply by dividing the VA rating by your incoming line voltage. In my case 300VA / 240V = 1.25A. This is the simple way to calculate the value needed for your mains fuse.

    I actually chose a 400VA transformer and kinda by mistake. Rather than using 7 amps as my top current limit, I accidentally did all the sums for my power supply with 10 peak amps instead. No harm done, although 400VA is where the mitigation of inrush current starts to require more care.
     
    Saint Johnny likes this.
  16. fully_articulated

    fully_articulated Forum Resident Thread Starter

    Time to prepare the transformer wiring by twisting them into their respective pairs, the white being the primary (input) and red & black the secondaries (output). I'm not sure if it's easy to observe in this photo but the primary and secondary wire thicknesses are quite different - I imagine due to the input side drawing only a fraction of the 11.1A rated output current.

    [​IMG]

    I've been using a coarsely cut rubber circle as an interface between the tranny and its bracket. Once I receive the compass cutter I ordered I can make a nicer looking one. This rubber reeks too, making me feel a bit ill. Might try submersing it in a vinegar & baking soda solution.

    Once back in place the termination of the wires began.

    [​IMG]

    I wanted to make this power supply as modular as I could, each section able to be replaced, repaired, or upgraded without too much trouble. This meant a liberal use of tabs, rings, and screw connectors.

    [​IMG]

    The bend radius of the secondary wires is tighter than I would have liked but it doesn't look like it'll cause a problem. The likelihood of these terminals falling out is minimal - their grip on the tab is very tight - but I've insulated the full length of them anyway as a precaution.

    From the front:

    [​IMG]
     
  17. Shiver

    Shiver Forum Resident

    Location:
    UK
    Cool thread.

    Had a M2 clone professionally built for me earlier this year. For shame I didn't have the knowledge to fully understand all the update build photos he kindly kept providing, but appreciated them and it's neatness nonetheless.

    It's an amazing amp, and I couldn't have more respect for NP's approach and ethos.
     
    Saint Johnny and SandAndGlass like this.
  18. Fdee

    Fdee Well-Known Member

    Location:
    Los Angeles, CA
    Looking good! I love the design of the daughter boards for the rectifier and TIO.
     
  19. Very interesting & thanks for your excellent pics, tech specs & layout info. It seems like a great idea to be able to tweak several Nelson designs too. I would have thought a seperate outboard power supply with umbilical an easier implementation but DIY always reveals unique solutions & learning experience especially when it is shared so well.
     
  20. fully_articulated

    fully_articulated Forum Resident Thread Starter

    Thanks for the reply Shiver. The M2 is very popular and its owners seem to love it. It's going to be hard to decide whether to try it or the Aleph J next. Feel free to post some of those pics here if you like.

    I have to admit, as much as I enjoy the learning process for something like this, what initially kicked it into gear was my jealously of some of the beautiful custom amps I'd seen from better DIYers!

    Agree. I can't judge a designer on their technical skills, but one of the things I do appreciate is pragmatism. And Pass has that certainly.
     
  21. fully_articulated

    fully_articulated Forum Resident Thread Starter

    Cheers! Cross fingers that it doesn't melt...
     
  22. fully_articulated

    fully_articulated Forum Resident Thread Starter

    You're not wrong. I did consider separate boxes for the PSU and amp, but I simply don't have the space. Thank you for your kind words.
     
    Last edited: Jul 22, 2020
    2channelforever likes this.
  23. Looking forward to your christening then & with so many future builds to test maybe we'll all see 2nd & 3rd platforms too.
     
    fully_articulated likes this.
  24. fully_articulated

    fully_articulated Forum Resident Thread Starter

    Deconstructing the First Watt DIY power supply - Part 3A

    Sorry, it's gonna be another text laden post. And it may be a bit sketchy because it's a subject I'm not wholly clear on, but let's try anyway.

    Where's your filter?

    A power supply really only has one job - as much as possible deliver ripple free voltage - and for that purpose the filter section is truly where the rubber meets the road. Once converted from AC, the DC output from the rectifiers is still very lumpy since all that's happened is the negative portion of the AC signal has been flipped to positive.

    Hopefully this handy chart will provide a clearer explanation.

    [​IMG]

    At the point just after rectification the ripple appears to be pretty much total (100% ripple-y), whereas in a perfect example we'd like it to be zero. A flat clean line of DC.

    The job of combating ripple falls firstly to the capacitor. A capacitor is like a very short-term battery, charging quickly and then at the part of the cycle the DC voltage drops it begins discharging, smoothing the intense ripple. This process can also be seen in the above graphic.

    CRC for you 'n' me?

    The First Watt PSU contains a CRC filter. Capacitance-Resistance-Capacitance. Capacitors C1 to C8 and resistors R1 to R8 form two CRC filters as seen in the schematic below. Don't worry if it looks like hieroglyphics, it'll be fine.

    [​IMG]

    If using the diyaudio universal power supply PCB and the standard component values, both the positive & negative rails will have two 15000uF caps, then four 0.47 ohm resistors in parallel, then two more 15000uF caps. Each group of caps is in parallel too and is perceived electrically as the sum of the two, in this case as 30000uF.

    Let's use the equations on this page https://www.electroschematics.com/capacitor-input-filter-calculation to analyse the effect of the 30000uF input capacitance on that ripple.

    Ripple = capacitor supply per cycle x current / capacitance x frequency

    Notes: The below sums incorporate the 70% discharge-per-cycle figure referenced in that article. The frequency is 100Hz because the bridge rectifiers double the line frequency (which in my country is 50Hz). Capacitance value is in Farads.

    Solving for 25W into 8 ohms:
    0.7 x 1.77A / 0.03F x 100Hz = 413mV

    Solving for 100W into 2 ohms:
    0.7 x 7A / 0.03F x 100Hz = 1.63V

    So it looks like as current increases so does ripple, but these figures seem very good especially considering this is just the first-stage capacitors at work. And I may be crazy but it also suggests that the secondary voltage plays little role in resulting ripple.

    Adding more capacitance at this point would lower the ripple further, but not as significantly, and the resulting ripple would still be an angry sawtooth shape. To take care of that we need the low-pass filter created by the remaining resistors and capacitors - the RC part of the CRC filter. See next post.
     
    Rick Bartlett likes this.
  25. fully_articulated

    fully_articulated Forum Resident Thread Starter

    Deconstructing the First Watt DIY power supply - Part 3B

    CRC vs CLC: Fight!

    Once the majority of the ripple has been squashed by the input capacitors, making that DC nice and smooth is the job of the low-pass RC filter. It may be best to leave the technicalities up to this article Power Supplies, Filter Circuits

    I confess I never cared, nor sought to find out, how much capacitance would be good in the second part of the filter. The First Watt supply uses the same 30000uF as the input does so that was good enough for me. I enjoy the symmetry if nothing else. But while researching for this project my eye was caught by the CRC filter's more elegant and effective sibling. CLC.

    The difference between the two can be demonstrated in simulation software like PSUD2. First let's look at the CRC filter.

    [​IMG]

    The red sawtooth wave is the voltage exiting the input capacitor bank. It may look particularly nasty but at this point in the simulation it is only 284 millivolts. The RC filter takes that down to 45 millivolts, and softens the sharp points of the original wave as the green trace shows.

    This should be very close to what is actually happening in the standard First Watt power supply. But, replace the R (resistance) with L (inductance) and the following occurs.

    [​IMG]

    Ripple out of the first cap bank is largely similar to the previous example at 301mV, but post LC filter it is down to a measly 2.3 millivolt, an improvement of over 40 millivolt on the RC filter and with an inductor of only 5 millihenries. If the unachievable target for ripple is 0, then the softly undulating green wave of the resulting DC voltage comes pleasingly close.

    One may ask then - why use an RC filter at all when the superior LC is available to us?

    [​IMG]

    Firstly, these things are enormous. Rated the aforementioned 5mH at 10A, they were the largest I could fit into the space. Higher inductance ratings will come at the cost of an increase in size. Also price. Certainly, using a few resistors instead is a small price to pay for reclaiming the space within ones chassis.

    Secondly, when electricity is run through inductors they produce a magnetic field, much like how a length of wire coiled around an iron bar becomes an electromagnet when power is applied. This may cause electromagnetic interference to be sprayed throughout the case. This is why I have decided to mount them within the steel transformer covers, in the hope of suppressing this problem as much as possible.
     
    timind and SandAndGlass like this.

Share This Page