Saturday, 14 October 2017

AD8307 Response to Sin and Square Waves

During the engineering phase of a project, there most likely, or should be a time where a question will arise that will take you off in another direction. Latest examples of this are the characterizing of amplifier outputs at RF frequencies, AD8307 behavior with sin and square waves, and RF measurement techniques. All of which can involve a considerable amount of research time that adds to the project development phase. It's not necessarily a bad thing, and one of the things I like about the hobby.

Another thing that can happen is while doing this research, new projects come to mind as I contemplate new found knowledge.  The SNA project has brought in for consideration a VNA. Learning more about complex impedance's and how to measure them has now got me thinking about a project involving the AD8302 board I bought on spec a while ago.  Now that I have amplitude and phase I can do some interesting calculations.  Except now I have to learn how to do polar charts and of course the intricacies of the 8302. Again all fun.

I made some comparisons between two AD8307 chips.  One smd sourced from eBay and the other a dip package from Digikey.  The Digikey part was sourced about 5 years ago.  I ran the test using both Sin and Sq waves.

Interesting results. In that the the package type overall did not really differ and I'll come back to that. There is enough of a difference between the sin and sq wave that I think you would need to  consider the signal type when making measurements.  Especially at the lower frequencies.

The sq wave results are very linear and the sin wave not as good but still not bad at all.  Considering that the 8307 output is 25mv/db,  I'd say you definitely would need to be aware of what you are measuring when taking the readings.  Here are the numbers for the spread between the sin and sq waves:
71 62 51 40 29 17 5  


Monday, 2 October 2017

A New Toy

I was able to obtain some great additions for the shack over the weekend. Helping someone clean up their spaces of no longer used equipment and find new homes for it was time well spent.

I was able to obtain a HP 3595B Spectrum Analyzer.  Great addition to the shack but unfortunately it will only go up to 40MHz.  However it will still be a useful piece of test equipment and fills a hole in that aspect that I didn't previously have.  I can at least see the 1st harmonic for most of the hf band and more than that for the lower frequencies. 

I still have plans to build a spectrum analyzer so it will be interesting to see some comparisons to the big expensive unit.

As well I picked up a bench meter, cable organizers, hand tools, and a resistor capacitor substitution box. I sure could have used that last week when doing my impedance characterization tests.

Tuesday, 26 September 2017

Almost finished but not really.

I think I can say that the lpf/mar6 circuit is finished. All’s left to do is documentation.  Making some more changes so the prototype has all the components that will be in the final product, had me add in an 8volt regulator.  Of course when I started this board it had not been decided as to what voltages would be available so I went ahead and fed the MAR-6 with 12v.

To accommodate the regulator, I had to think up some way of fitting it in.  No room on the board that was easy so I came up with the idea to insert the regulator in-between molex connectors. I soldered the pins onto the regulator and inserted the load side into the molex connector mounted on the pcb.  The input of the regulator in soldered directly to a connector and the original 12v feed plugs into that.  Worked out pretty nicely and I didn’t short anything out. Pictures below detailing this.

Interestingly the most important thing that came out of this session was making rf measurements.  This topic rears its head every one and a while and this time I took the opportunity to dig into it a little more.

Making measurements during this project became very frustrating. Noisy signals along with measurements that did not agree on repeated tests been the primary cause.  Hand/body placement or nudging the probe was enough to drastically alter the readings.

Of course it pretty much came down to the ground lead on the oscilloscope. Reading up on the subject and most importantly, viewing Alan’s, W2AEW, video on this was most informative.  Alan describes a spring or a wire wrapped around the ground contact on the probe as an effective way to minimize ground lead induced problems.  Going through the bag of items that came with my new scope revealed such a device. 

So as much as I hated the thought of more testing, I was curious to see the difference in making measurements with it.  Big difference. The signals were now much clearer with less noise, making those measurements easier.  Only problem now is figuring out a way to make a probe socket to mount on prototype boards to make the measurements easy.  So here we are again, a project driving the need for another project.

 73, Peter

Saturday, 16 September 2017

The Saga with AD9850

Just when you think you have covered all the bases and have finally got it right, along comes the inevitable. If you’re lucky the circuit works as planned, works but not quite right, and in some cases you let the smoke out. 

After a few rounds of testing and trying hard to not pay attention to that little voice in the back of your head, you unplug the power and take a closer look at the board, datasheets, and the schematic.  In the case of my lpf/mar-6 board, a missing cap, a resistor that should not be there, and another missing capacitor.  So the cure was cutting a trace, adding stuff, removing stuff, and voila, it works.
It’s amazing to me the feeling you get when something you build from scratch works and to go from the frustrations when it doesn’t work to when you fix it, even more.  Mind you it would be nice once and a while to have a circuit work right out of the gate.

It’s an interesting project where on one hand it first appears to be a simple project.  Combine an lpf circuit with an amplifier and away you go.  So where did this get complicated?

1, Board layout. Need to go through 3 or 4 revisions at least before getting it close enough to actually etch.

2, Components you thought you had on hand, you don’t, or are of the wrong footprint.  Now you have to beg or borrow, wait a month for that slow boat, or pay some hefty shipping charges.

3, Errors you make due to stupidity or a lack of knowledge. The latter being ok as it is all about learning. The former, well make up an excuse to explain to your buddies.

4, We now have the option of simulating circuits. Confirming that the circuit at least has the chance of behaving the way we expect or at least get you in the ballpark.  LT spice to the rescue. However, there is a learning curve that is well worth the time spent in my opinion.

5, By now you are in so deep with the prototype, it can be hard to let go of it.  Recognizing when you can continue with the present board or do start over.

6, Working with a team on a project. Unquestionably a great way to go about a project. I would never have learned so much on my own as I do being part of a team.  Individuals bring different strengths and knowledge to the group, and besides, it’s way more fun. So why do I have it listed under complications? Well, plainly stated, information overload. It’s all inversely proportional of course to the amount of time one has. The better organized you are the better you can track, remember, and implement those new ideas and suggestions.  Who would have thought that I can be busier at my hobby than I was at work before I retired? 

For my latest board, going to give it another go. Couple more modifications and some more testing that hopefully turns into a proof of concept.   Pictured below, shows the board in its present state.  The fixed resistors for biasing the MAR6 were removed and a pot substituted. This will give me better control over setting the current. Missing cap added in and another one still to be added. The thru hole cap visible has a trace broken below it to allow for circuit changes required due to oversights on my behalf…. To put in nicely. 

Friday, 8 September 2017

AD9850 LPF Redesign and Amplifier Board

Almost finished the AD9850 LPF/amplifier board.  Just a couple of parts to left to install and they are being mailed to me. Hopefully Canada Post can beat the slow boat from China and I’ll get them by Tuesday next week.

When designing the board, there were 2 options for selecting a footprint for components. Example, 1210 or 1210 handsolder.  I chose handsolder but I really don’t know exactly the differences. I assumed the pads are a little larger, but the tolerance between the contact area on the device and nearby ground spacing is tight. As well the distance between the solder pads seems to be a little on the far side.  Not much room for error.  Anyway no big deal as I check for shorts every time I place a component and a good thing I do that as I do get shorts. 

My boards are on the smaller side of things lately and I’ve noticed that my printer has an easier time with the print density.  Not as many pin holes on the ground planes.

One change I did this time round was to use a pin header for test points instead of individual test pins located throughout the board.  Since this board is not overly complex, I was able to group 5 pins together and this gives a solid support so badly needed.  My previous methods didn’t last too long as the constant attachment of probes eventually ripped up the pad that the tp was soldered to.  Even enlarging the pad didn’t help much.

I did manage to damage one trace and needed to put a jumper in to repair. It was the 2nd component I installed so I almost was going to scrap the board and re etch another but went ahead and carried on.  It’s a prototype board and not the finished product.

I had to change the 2 15pf caps in the LPF to 18pf as I was out of the 15’s. Fortunately LT Spice basically shows not much of a change in the characteristics of the curve so all’s good with that.
I have an Altoids tin available, almost out of them, so I’ll install the board in that to minimize outside influences. 

That’s it for this board until next week unless the other part of the design team points out an oversight. Well actually that has already happened but I was well into this board so I ignored it…for now.  Next version of the board will have a couple of attenuator pads etched in. One ahead of the filter and the other right after.  Thank you Dave.