Willow II


2013-10-18 2:42 am
I could not contain the excitement any longer and had to share the fruits of my labours over the last 7 months.
Willow II Born.jpg
It all started with me contacting Robert Nance Dee the designer of my previous build consisting of his PI and Willow I.
He mentioned he had been working on a new design for a preamp the Willow II and it was due to be published in the February 2016 edition of Audioxpress magazine. He then generously asked if I would like to be a tester for him and have a go at building it! When a great designer asks that you cannot say no.
I had no idea what I was getting into and Robert loving minimalist builds, this was all SMD!
He kindly supplied me the PCB's with a BOM and off I went ordering all the parts I was so excited. I'm going to do a complete build of this preamp later on but just to create interest and show what is possible.
Willow II Born 2.jpg
And yes it is Arduino powered!


2013-10-18 2:42 am
Thanks GOR3 and vzaichenko for the compliments.

OK as promised full details of the build it has taken me 7 months to get to this far.
I suggest you get a copy of Roberts feature article in February 2016 AudioXpess magazine to get the finer detail. As the article on the front said it best this is an update of the previous Willow I preamp and attenuator PI.
So armed with the BOM I set to and ordered everything I would need lots of SMD devices. Once they arrived some of them were so small the only way for me to see them was by magnifying glass. I had received two small boards and a larger one (main PCB).
The heart of the preamp is an ATiny MCU located just to the left of the displays that drives both the 3 seven segment displays and the I2C bus feeding the dual digital pot AD5282.
Behind the vice is the display and in the vice is the selector board using hall effect diodes and 4 neodinium magnets to drive the fets that switch the relay selection. Here you can see the PSU and the main preamp board.
IMG_20151107_222401_hdr (1280x1280).jpg IMG_20151107_222423_hdr (1280x1280).jpg
Here you can see the FET front end the good old PMBF4393 which is fed by the AD5282. The digital pot chip is at the limit of my soldering abilities.
IMG_20151108_182815_hdr (1024x1024).jpg
So the question everyone should ask themselves are SMD devices worth it? All I can say is yes and once you have mastered the technique it is so much faster and accurate than through hole components. The trick is the solder paste that you apply to the PCB pads beforehand with a syringe. This has two functions one to seat the component so it does not slide around too much and the second to melt on contact with the soldering iron tip or heat gun. Holding the component in tweezers can be a bit tricky but once you have done a few it is easy.
IMG_20151215_195105 (1280x1280).jpg
Here is the wonderful case off Ebay somewhere in China very well made and anodized silver front plate. Here you can see I’m laying out where things will go and the size and location of the cutouts needed.
IMG_20160124_085910_hdr (1024x576).jpg
I am fortunate to have access to a milling machine at work so one evening managed to complete the front panel.
IMG_20151229_135635_hdr (1280x960).jpg
The main PCB has PCB mounted RCA connectors so I simply measured it up and drilled away. The case already had a cutout for an IEC socket and fuse holder!
IMG_20151230_230058_hdr (1280x960).jpg IMG_20160124_184351_hdr (1280x720).jpg
Laying out the power supply board as you can see the main PCB is self-supporting using the RCA connector fittings. You will note the PSU has two transformers and my front panel did not have an on off switch. As I was building this preamp I like certain minimalist ideas like Robert so I had decided that the volume control was going to be done via an encoder and one can obtain encoders that have a momentary switch built in by pushing the shaft. So I built a simple latching circuit so the small 3VA transformer is on all the time and the NE555 timer provides a latching function and has the capability to drive a relay direct which in turn supplies 120Vac to the main transformer.

This next circuit board on the right was going to be the wireless remote control as I have previously done for the Albert shown elsewhere on this forum. I was going to use discrete components together with a cheap transceiver pair from china that are excellent value for money.
IMG_20160104_075330_hdr (1024x768).jpg
So before I started to build the remote control I assembled it completely and powered it up, the display lit up momentarily and then went off. I immediately thought something was wrong and switched off. It turns out that Robert had designed the display to go off after movement.
Since I have no on off lamp to tell me the unit is powered up I did not like this and I reported this back to Robert who said he’d make changes to it before the final version went to print.
One thing being asked to test something the urgency felt by me to get this built quickly so I could report back my findings so any changes can be made before it is unleashed on the public.
So this is audio so the burning question is how does it sound? I had recently just completed the F6 amp and was eager to plug it in and sit back.
I sat back and was greeted with total silence the quiet power supply lived up to the design! But when turning the volume encoder up this weird buzzing sound occurred and then as soon as the LED’s went off it stopped but totally ruined the experience for me. Once a volume level has been set the sound is fantastic deep bass clear treble and wide sound stage.
So after the initial disappointment I then started to do some serious listening with headphones on together with a good cup of tea. On placing my cup down on the table next to the Willow II I heard my cup! I thought I was going mad but when I tapped the case it was microphonic! I then removed the cover and started to tap the main preamp PCB as this has the finest and smallest components so maybe I had got a whisker of solder somewhere but no could not make it make a sound. Maybe my latching power supply board was the culprit but no it too was silent.
Then moving on the Willow II power board the merest scratch or nail across the board could be heard.
So then removing the power supply board from the case and putting my scope on it with no load on it I could generate large swings in the +- 15 volt rail and 38 volt rail. I cleaned the board remade many of the connections but could not find the fault. I emailed Robert and told him the findings he immediately sent me anew PCB thinking is was some short or a faulty via in the PCB.

I remade the board and with great excitement put it back in the case and powered up and to my dismay it exhibited the same symptoms as before. I was completely stumped I had used the exact components that was on the BOM so what could it be?
I then decided to make my own power supply with my parts box good old 7805, 7905,7812,7815,7915 and a good old 317 to see if there was somehow some weird interference caused by the digital bus or chips on the main PCB. I powered up my creation and it worked no microphonics.

Completely stumped by this I then started to look at the power supply in detail checking everything I changed out a couple of LT3080 and LT 3082 still no good the bridge rectifier next but still nothing. Bear in mind these are surface mounted devices and are not the easiest to remove so this was another learning exercise. I was completely at a loss and so I checked the resistors one by one and nothing they all read correctly!
Then I turned to the capacitors and low and behold it was these ceramic capacitors especially the ones on the reference pin of the regulators that provide some decoupling at the top of very high value resistors 750K and upwards. I had never come across anything like this before so I searched Google and sure enough I came across this article.

This is exactly what I was seeing.

Snapshot 1 (2016-05-17 10-06 PM).png

This is a screen shot of the +15volt rail 20mV per division on the 38volt rail it was over 2 volt swing with just my finger nail touching it.
I replaced them with film type not ceramic and the problem solved.

End of Part One


2013-10-18 2:42 am
Part Two.
Robert had made me very curious regarding what he had managed to do with the ATiny MCU being able to read an encoder, drive the digital display and then via the I2C bus control a digital pot I was in awe.

I wondered what an ATiny was and how did people program these so Google to the rescue and I started to see mention of this chip/board Arduino. More research turned up this whole community out there writing programs and creating amazing things with these cheap Chinese boards called UNO’s. I started to read as much as I could and started to realise that with a bit of knowledge maybe I could use these to control the Willow II. Then on Ebay saw these really nice OLED displays and thought these looked way better than the seven segment ones on the Willow. Since these things were so cheap I thought why not at least give them a try and if it proves too difficult then I will make do with what I have just built.
I ordered a couple of displays UNO’s and a few other boards and chips to experiment with. Managed to download a 7 hour video tutorial on learning the basic programming of the Arduino and UNO. I started out to see if I could make a selector and light LED’s and then learned how to read an encoder. Now the test was to see if I could read the encoder and display it on the OLED display that I had bought.
Here is my first attempt.
IMG_20160325_163307 (1280x720).jpg

It worked!
Learning Arduino is a very frustrating exercise as the semi colons and curly brackets are a nightmare get one in the wrong place changes the whole of the program, fortunately the compiler tells you where you have gone wrong but not exactly what it is.
Since this was going to have wireless remote control I needed to see this display from a distance and so I rewrote the program to change to larger digits and letters when the remote is keyed. Having got that sorted out I felt the display was not large enough (read optically challenged) so another one was purchased and it is what you saw in the final version.

IMG_20160606_073019 (1280x720).jpg

The veroboard on the right is my wireless remote control shield plugged into the UNO underneath. It was cheaper and more convenient to place the UNO in the case than purchase a Mega328 CPU and the associated components plus I get to program and debug the unit with a simple usb cable.
So now having mastered the display I turned my sights on the I2C bus control. Once you learn Arduino and it is open source many kind people publish their code for anyone to use, it is easy to download a library to do the function you need.
I needed to communicate the Volume count from the encoder and send it to the digital chip via the I2C bus and someone has already written it for me so I simply drop it into my program and voila. Well not so voila it only controlled the volume on one channel not both.
After studying the data sheet over and over again I must have missed something and sure enough the dual chip has to be addressed differently and once that was done it worked and I’m listening to it as we speak!
The knobs are just temporary and will be replaced by some nice machined aluminum ones.
Off now to build a CNC using the Arduino to make circuit boards for more audio gear!

Thank you for reading.


2014-05-12 7:47 am
PMBF4393 is obsolete now. What jfet we may use in its place in this simple design, of single jfet providing all the gain.
Instead of the output opamp buffer, can we use common drain stage with same jfet for output buffer, or directly any other power buffer amplifiers like MoFo to make a integrated amplifier?
This is a screen shot of the +15volt rail 20mV per division on the 38volt rail it was over 2 volt swing with just my finger nail touching it.
I replaced them with film type not ceramic and the problem solved.
Yes, such ceramics are not used for sensitive analog, or at least not without electrolytics in parallel. 'Scope front-ends are classic examples of the same microphony - I just tapped the unconnected BNC socket of my scope lightly on the 20mV/div setting to get this: