Building a Pedja Rogic Buffer
Building a Pedja Rogic Buffer
- Category: Amplifier
- Published: Sunday, 12 June 2011 20:00
- Written by Justblair
This article covers the populating of the PCB
The design for the buffer comes from Pedja Rogic. A gentleman who has contributed much to the DIY audio community.
He now runs his own business Audial.
From his website you can see the full schematic with details on its operation.
In my design, I chose not to attempt to cram all of Pedja's design on the board. For a start I am not using the LM1875 initially in my amplifier that this is destined for. Secondly, a 4.7uf Polypropylene capacitor is a fairly large device.
Instead, the DC coupling capacitor is going to be used to link the buffer to a Tripath based amplifier.
Basically then, my design includes everything that you see to the left of the 4.7uf capacitor is what is included on my board. in addition to this, my design includes a 1 meg resistor between the base of the 2sk17-bl and ground. This serves the purpose of referencing the 2sk170BL base to ground in the absence of a volume pot, or if the volume pot fails.
This was a suggestion by Pedja himself, who helped me in the PCB design by pointing out a couple of issues during the design stage. I have to say a heartfelt thank you to Pedja, who not only took time to answer my emails, gifted the internet with his design and saved me from my mistakes. Bear in mind Pedja has his own HiFi company, so giving his advice for free is a generous gesture.
The company I chose to outsource to was Futurlec. I shopped around, and for the volume that I was interested in Futurlec came out the best for price. Options wise I went for everything. The Board, Silkscreen and Solder mask. I saved a little money as the PCB is a single sided design. Futurlec worked from the Gerber files that I produced from PCB.
This is the resulting PCB. It is compact. This was quite deliberate. Asides from a small cost saving, this allowed me to use the smallest possible trace sizes. Trace lengths on a PCB have an inherent resistance, albeit very small and usually more a problem on high frequency circuits. The design I used mounts components on both sides of the PCB to keep these track lengths as small as possible.
Component wise I opted for some high quality items. Particularly so was in the capacitance. The 1uf Capacitor that feeds the Cascode in the design is a Sonicap Gen I. I have seen other builders use a variety of polypropylene capacitors in this position, but usually not boutique items. The Sonicap has a reputation as being a high quality audio capacitor at reasonable prices. At the 1uf value it was not going to break the bank. Its shape also happened to fit my design ideas.
I also designed in space for a Vishay MK1837 bypass capacitor. This is an addition to the 1uf capacitor that Pedja designed in. The reason behind it was that on Humblehomemade HiFi this capacitor was reviewed and in the signal line, improved the top end of the Sonicap capacitor.
In Pedja's circuit the IRF530 supplies additional current to the 2SK170 using the output from the 2SK170 via this capacitor. Using a bypass here is almost definitely over kill, but it will be interesting to see if the bypass changes the sound.
Other steps that I took to ensure the highest build quality possible were:
- hand matched the IRF3530, 2SK170 and even the J310’s for stereo pairs.
- Used 25 turn precision cemet resistors to allow precise setting of the current limitation of the J310 to 6ma
- Used Rubycon ZA caps in the power regulator stage alongside smaller polyester capacitors.
The power regulation could have been improved upon. I Used lm7815 and LM7915 +/-15v voltage regulators to supply the boards. The ripple rejection of these can be bettered using say LM317’s or discrete regulators. However using the LM7815 and LM7915 allowed the power to be placed very near the circuit and board size to be kept compact. I intend using this circuit after a first stage of power regulation, so hope to minimise ripple at this stage.
Populating the board.
Everyone has their favourite methods of populating a circuit board. What I find easiest is to begin with components that have the lowest relief from the board surface. In the case of this circuit that is the 1% Metal film resistors. I began with them.
The soldering technique is important here. Some tips that I can suggest:
Always start clean, especially with a board that was made a while ago. Solder does not stick very well to corroded surfaces. If using a corroded surface, a fibreglass pen is a useful tool. The fibreglass can be used to rub gently away the corrosion on the PCB. Don’t neglect the component legs themselves.
Place your component in the board, with the legs in their respective holes. Turn the board over and trim the legs off before you solder. This is easier if you work from the lowest components to the highest as the surface of a table can be used to push the component flush to the board. The advantages of trimming the legs beforehand are two fold. One you get a neater more professional looking join, more importantly, when the solder touches the component leg, it has to heat up less errant metal. This means that you apply heat for less time and your component itself heats up less. To make the join, apply the iron to the metal of the pad and component leg and bring the solder to the two rather than to the soldering iron itself.
You can see when you have achieved a good solder joint. The solder should flow easily to the metal of the component leg and the solder pad of the board..
There should be no bulging of the solder at the pad side, more a cone shape created by the solder.
Once the resistors were soldered, I moved on to solder the J310 and 2Sk170 in place. These components are most susceptible to heat damage. A tip that you can use when soldering very susceptible components is to clip a miniature crocodile clip or a pair of self closing tweezers to the component leg on the component side of the board. This acts as a heat sink preventing some of the heat reaching the component body. Make the solder joint as quickly as possible.Also a good idea is to allow some cooling time between soldering the different legs of the same component. This will prevent the component being heated unnecessarily.
The next highest components were the Vishay 1837, the trim pot and the power supply capacitors. I began with the Vishay 1837 and the Bournes 1836 precision trim-pot.
I installed also the vertical 25 turn precision trimmer in place. Once complete this trimmer will be used to tune the amount of current that the J310 will pass to 6ma. I left the legs long on the trimmer, simply to give me an easy to reach test point on my PCB.
With the variable resistor and Vishay Capacitor in place I then went on to solder in the Wima MKS02 and the Rubycon ZA capacitors. These are part of the power supply section. The Wima MKS02 Capacitors have no polarity, so it does not matter which way round they are soldered.
The Rubycon ZA on the other hand have to be soldered in the correct way round or else their will be a sometimes spectacular failure of the component. You can see in the picture the correct orientation of the caps in the board. The negative leg of the capacitor faces in the way. The white stripe on the body of the capacitor indicates the negative side.
If you are working out for yourself which way round a capacitor goes, the negative leg goes to the most negative side. On a positive voltage this is easy, the negative of the capacitor attaches to ground. On a negative voltage rail, it is the positive side of the capacitor that attaches to ground.
With the capacitors in place, I then installed the power supply connection block.
Next I installed the Molex connectors that connect the signal line and signal ground to the circuit. To save space and reduce the PCB tracks, I cheated slightly in my design here. The right handed Molex connectors are actually designed to be mounted on the very edge of the board. The issue this causes is that if soldered in with their bases flush to the board, they do not allow enough clearance for the plugs to fit in. The solution is simple.
Prior to soldering in the board, fit one of their plugs onto the header… I used self closing tweezers to clamp the header and plug in place on the board while I soldered. You can see that when done, the header has a mm of clearance between its plastic base and the PCB.
Second Last on my list of items to solder were the TO220 packages, namely the IRF530, LM7915 and LM7815 as these are the tallest items on the board. Again the legs were trimmed short to aid with the soldering.
Finally the 1uf capacitor is mounted on the rear of the board and the board mounting screws and spacers are added to allow board mounting.
Here you can see the complete circuit board. It is ready to test. Before doing so, set the variable resistor so that it measures 270-370R. The legs were left untrimmed are the test points for this.
Test the board first of all by plugging in the power. First thing to check is that the voltages are correct on the power supply rails. These should be very close to +/-15v.
Assuming this is ok, next job is to set the variable resistor so that it allows the J310 to pass 6ma. This is going to take a little calculation in order to derive the current. We shall use Ohms law to derive this.
Once this is all set have a quick check to ensure that nothing is running hot before getting ready to test the buffer works. I used my oscilloscope on the output, and fed the buffer with a sine wave. I generated the sine wave using my laptop. There are various software packages that will give you a test sine. Alternatively download a file with a recorded wave and play it through a media player. I have even used an iPod in the past to play one of these files.
Finally connect the buffer to your chosen amplifier stage. Be careful to ensure that the ground of the buffer power supply is tied to your signal ground somewhere in the chain or you may get some unpleasant and noisy results.
Working on the quality aspect of this project, I decided not to manufacture my own PCB. In the past I have manufactured PCB’s using the photo resist method. Whilst it is possible to make perfectly useable boards in this manner, I don’t have the skill or technology to silk screen the boards. Another factor that swayed me to outsourcing the PCB manufacture was that I intend making several of these buffers. When you get to even small volumes of a PCB, manufacturing becomes cheaper both in financial terms and also in terms of free time consumed.