3286	2009-07-01 11:38:27	Ashhar Farhan	The importance Q in a matching network
Message	Date	From	Subject
The normalized pi-network is a standard in our toolbox. Often, at the beginning of a design, we have to specify the Q for the network. what is the significance of this choice? why are most of our power amplifier matching networks designed for an unusually low Q? For instance, in a 7 MHz amplifier, our interest is primarily in the region of 7.0 to 7.2 MHz, a bandwidth of 200 KHz, this point to a required Q of 35. But the networks are often designed with Q-Loaded of less than 5. I am raising this issue because I am struggling to understand what happens when a standard pi network is used to tune across two bands (like 7 and 14) by varying the capacitors (keeping the same inductor). I can visualize that the two capacitors form a divider across the inductor thereby achieving the impedance transformation. Though the amateur literature is full of admonitions to keep the Q low, there is a lack of clarity on why this is required. Can someone guide me? - farhan
3287	2009-07-01 19:30:33	Russell Shaw	Re: The importance Q in a matching network
Ashhar Farhan wrote: > The normalized pi-network is a standard in our toolbox. Often, at the > beginning of a design, we have to specify the Q for the network. what > is the significance of this choice? why are most of our power > amplifier matching networks designed for an unusually low Q? > > For instance, in a 7 MHz amplifier, our interest is primarily in the > region of 7.0 to 7.2 MHz, a bandwidth of 200 KHz, this point to a > required Q of 35. But the networks are often designed with Q-Loaded of > less than 5. > > I am raising this issue because I am struggling to understand what > happens when a standard pi network is used to tune across two bands > (like 7 and 14) by varying the capacitors (keeping the same inductor). > I can visualize that the two capacitors form a divider across the > inductor thereby achieving the impedance transformation. Though the > amateur literature is full of admonitions to keep the Q low, there is > a lack of clarity on why this is required. Can someone guide me? The source and load deliver and consume a certain level of current. The current circulating in the pi network increases with Q, so Q=35 means 35x more current will be circulating (assuming source and load are resistive, but in practice, they're not 100%). To achieve that, the pi components need to have a reactance around 1/35th of the source and load (approx, because Zsource and Zload can be quite different). This means that for the pi to have low loss, the pi components need to have extremely low loss, and very low impedances, making it more unrealistic. Similar applies to Tee networks, except that you'll get ~35x higher voltages and 35x higher reactances than normal.
3288	2009-07-01 22:22:55	Ashhar Farhan	Re: The importance Q in a matching network
russell, thanks for the explanation. it is a little counter intuitive to shoot for lower Q for lesser losses. but i understand that we are talking about loaded Q and not the unloaded Q. the less the energy in the resonator for lesser time, the less the losses. i guess this is what it means at a physical level. my investigation started with using a pi network (instead of the venerable L match) to tune the antenna. the currently fashionable 'balanced' balance line tuner uses two L networks to match line. i was considering using a two pi networks instead. this allows me to skip the variable inductors or alligator clips on an breadboard AMU (i have kids tearing around the shack at all times). - farhan
3289	2009-07-06 06:47:02	Stan	Re: The importance Q in a matching network
Farhan, I suspect the reason we do not see the Pi network as much today has more to do with the changes in the design from vaccum tube to solid state transmitters. The Pi network first came on the scene in the 1930s and often called the Collins tuner in the older radio books. It the 50's it became the major component in the transmitter since it was easier to build a tapped coil than the plug-in ones of the earlier era. The Pi network works best when the input impedance is a Hi-Z. As the input impedance is lowered the value of the input capacitor has to be increased to achieve a match. When you get to the 50 ohm input level the size and cost of the input capacitor becomes the major cost of building with a pi-network. Lowering the Q of the inductors allows lower values of input capacitor, it also has the effect one does not have to retune the finals when moving around the band. Retuning is not much fun with running a vaccum tube transmitter. Do a simple calculation of a Pi network for a 1000 ohm input impedance with 50 ohm output and using the same value of Q do a network for 50 ohm input and 50 ohm output. Watch the effect on the capacitor. Increase the Q and I believe the input capacitor will increase even more, however, it has been a few years since I designed a Pi network stage. They use to be a Pi network design on the web and you could put in freq and vary the Q, etc. think it might have been DX zone that had it. It was fun to play with; the current Pi tuner designer on the web will not run on my Vista computer so did not get a change to check it. So other factors in addition to a slight increase in loss. Application is different today. Stan AK0B
3290	2009-07-06 09:02:20	Ashhar Farhan	Re: The importance Q in a matching network
thanks stan and everybody else, the last few days have been of incredible education. the amazing thing is that everybody is right! there are a number of explanations and each one of them is so chestful of interesting insights. my immediate need was to build a balanced antenna tuner for my feed line without resorting to variable inductors. it occured to me that a pi network could do the trick. two paralleled pi networks would essentially become a lumped constants equivalent of the ladder line itself (two variable capacitors shunted across the input and output, fixed inductors as series element between the input output posts). as is usual, i first refered to EMRFD for the cookbook solutions and then I started digging deeper and deeper into it. right now i am lugging around my school days' 'hayt and kimberly' (Engineering Circuit Analysis) trying to understand complex impedances and complex frequencies with bob's 'caron' impedance bridge in hand. btw, here is an online calculator: http://bwrc.eecs.berkeley.edu/Research/RF/projects/60GHz/matching/ImpMatch.html The Q of a pi network with fixed inductor will vary with frequency. i found it best to bisect the pi network into two L networks whence the analysis becomes simpler. i wish there were a text that goes deeper than EMRFD, explaining the mathematical concepts like poles, zeros, complex planes, et al for the hams (not engineers). it would help us be understanding how our filters and close loop systems work. there probably is something like that already (IRFD is too theoretical and needs you to be familiar with fair amount of mathematics). - farhan
3293	2009-07-07 11:57:24	davidpnewkirk	Re: The importance Q in a matching network

3298	2009-07-08 11:57:40	Kevin Wheatley	Re: The importance Q in a matching network

3299	2009-07-08 21:34:25	Ashhar Farhan	Re: The importance Q in a matching network
wow, it is an amazing tutorial. i am reading it right now ... - farhan
3300	2009-07-09 14:44:24	gt25psi2002	Re: The importance Q in a matching network
Dear Sir, Thanks very much for the website. It is very informative (but scared to see the formulae!) I'll try to absorb it even though only 1% or 2% is my ability. Best regards, Amin > > I've been doing some research on this subject over the last few days, and have found a REALLY great site that explains things in understandable english (well most of it made sence). > Here's the link: > > http://www.g3ynh.info/zdocs/z_matcing/part_1.html > > 73's > Kevin - M0KHZ >