CD74HC4046A: Help defining resistors and capacitors for CD74HC4046AE PLL

Hey Danilo,

This E2E post does a good job of explaining the resistor and capacitors and the datasheet can help you find the values that you need : 

https://e2e.ti.com/support/logic-group/logic/f/logic-forum/998779/cd74hc4046a-frequncy-calculation

You can start by looking at figures 11 and 12 on page13 of the data sheet for R1 and C1. Vcc is 4.5 and 6V here so you'll be working somewhere in between. Say you select 1nF for the cap, it looks like a R1 value that would work with the frequency is between 220K and 330K. 

R3 and C2 are just low pass filter components. You'll want to select them such that your frequency cutoff includes the frequencies you're working with.

This post goes over how to solve this : 

https://e2e.ti.com/support/logic-group/logic/f/logic-forum/938637/cd74hc7046a-cd74hc7046a

Thanks,
Rami

Danilo,

Danilo A. said:

1) should you have used 10**6 and not 10**5 to get R1 and C1 at 1000kHz?

Sorry about that. You are correct. 10^6

Danilo A. said:

2) Could you take a minute and see what you get for a 180pF capacitor for R1?

This is using a center frequency of approx 1100 and 180pF. It's just an estimate since this is log scale. 
I'm getting somewhere between 22K and 30Kohms. It looks like you're falling pretty dead on the 22K and the 30K lines. For 5V I would estimate closer to 24-25Kohms would be a good starting point. This may take some fine tuning in actual implementation, however. 

Danilo A. said:

3) Can you provide a lead/lag filter example as shown in scha002 I can use to create the filter? My requirement is for a lead lag LP filter...scha002 does not go into finding the values...

This is in the app report

Here is a previous E2E where an example is done 
https://e2e.ti.com/support/logic-group/logic/f/logic-forum/938637/cd74hc7046a-cd74hc7046a/3466664#3466664

Following those steps here, and using just use the 180pF cap for C2 since we know they have that on hand already for this example. I get an R3 of approx 8Kohms.  

Thanks,
Rami

Hi Rami, Danilo,

This is a pretty complicated question to explain in detail without getting into the math behind it. In general, though, the following can be said: the standard R-C filter has a capture ratio that goes down as the loop bandwidth decreases. This isn't a big problem for AM radio since the carrier signal is nominally stable, but it might impose a challenge for FM radio where the capture range and the lock range must be similar, even across a narrow channel. Adding R4 adds a zero to the design and gives another parameter in the loop transfer function to play with, which can be used to keep the capture ratio high even with narrower loop bandwidths.

I'd recommend for AM radio just using the R3-C2 filter.

If you're still curious about the details, I recommend reading through the fairly brief but quite math-intensive Miniaturized RC Filters Using Phase-Locked Loop. (Moschytz, G.S.) which may be found in any number of places online, or the lengthier and equally math-intensive Phaselock Techniques (Gardner, Floyd) which is still in print today and can at least be previewed up to some of the relevant content on Google Books.

As far as designing with this thing, consider that you'll be tuning the AM capture bands with step-adjustments to R1 and/or C1 based on the choice of phase detector, and the loop bandwidth should be designed to be discriminating (small) enough that different AM channels are distinctly separated. The narrower the filter, the smaller the ripple will be, so the tighter the VCO frequency will be to the carrier, and consequently the better the AM demodulation will eventually be at the frequency multiplier. If the loop bandwidth is made too small, the capture range will be substantially reduced, which could make it hard or impossible to lock to the AM band you care about within component tolerances; furthermore, the smaller the bandwidth, the longer it will take to capture marginal frequencies as well. Figure out the combined component tolerance and how it will impact the frequency error for your AM signal, and work backward from the standard filter equation given in figure 46 until you get a capture range that's slightly larger than worst-case tolerance.

And of course, depending on the choice of phase detector, there may be issues such as locking to harmonics of the center frequency. As nice as it is for AM to have phase comparator 1 with automatic 90° offset at the divider output, the added narrowband input filter is rarely worth it. Since your VCO frequency can be made higher than your input frequency, you could arrange to always include a factor of four in the divide value, and manually generate a 90° phase shift without sacrificing the other benefits of phase comparator 2.

Regards,

Derek Payne

Danilo,

It isn't clear what you're asking here. It sounds to me like you have another device, 74HC4046N, that you're trying to replace with our device. However, our devices GPN is the CD74HC4046A and the OPN is the CD74HC4046AE. Not the 74HC4046AE as you keep referring to it. It's important to use the correct naming since a lot of companies including ours share very similar part names for this device and other devices across our portfolios.
I couldn't find a datasheet for the 74HC4046N. The best I could find was NXP's 74HC4046AN which appears to be obsolete. Is this what you're looking for? The CD74HC4046AE is a drop in replacement for this. If the values they initially had for the low pass filter worked why are we changing them? Has the CD74HC4046AE been tested with the values that already worked? It shouldn't require new values but as I said before, the best way to really see these sorts of things is to just check actually run it rather than do math and hope you're right. If they have the components and the chip, i'm not seeing why a test hasn't been done. However, this should just be an easy drop-in replacement. 

Danilo A. said:

is there another combination of current manufacture which I could substitute in to what I have selected instead?

I'm not sure who the current manufacturer is but it seems to me that these aren't already TI parts so I don't feel so eager to find competitor parts for you. If the intent here is to find TI replacements for the 74HC40103 and 74HC4060 then the CDHC40103-EP and the SN74HC4060 can be replacements that share the same functionality. Depending on the package, even P2P or drop-in replacements. 
To help you find the right replacements we do have a BOM & Cross reference tool on TI.com that can speed your process up. 

Danilo A. said:

My intent is to find a PLL and filters to replace the 74HC4046N...what is suggested replacement with current PLL manufacture then?

If this was the initial intent, It may have been more beneficial to start with this in the original question. The CD74HC4046AE would be a drop-in replacement for 74HC4046AN. 

Rami

Danilo,

You seem to be disregarding my notes here.

Rami Mooti1 said:

Has the CD74HC4046AE been tested with the values that already worked? It shouldn't require new values but as I said before, the best way to really see these sorts of things is to just check actually run it rather than do math and hope you're right. If they have the components and the chip, i'm not seeing why a test hasn't been done.

Has a test been conducted with the values that already worked with the old device?

For R1 and C1, as I mentioned above in this thread, the implications were initially that there was no offset. With an offset you'll have to use figures 27-32 not 11-15 as they used first.

Rami Mooti1 said:

The R2 resistor would need to be used. The way to find this value is also in the datasheet. Additionally in this section it tells how to find the offset frequency, R1 and C1. I won't go into finding these values since the datasheet explains how to do this and it's not clear whether they even need this. 

At this point R1, R2, and C1 would be found.

So given an approximate offset frequency of 500Khz, this would be ~5.7 ticks up using the log scaling) you would have the following for R1 and C1. Note here that the 4.5Vcc can be used for 5V Vcc as 4.5V is the worst case for 5V. 

If they're going to keep the 180pF capacitor as C1 it looks like R2 is between a couple of known values but the change there is linear so I would estimate ~130Kohm.
I would be more inclined to use a combination where the values fall on one of the lines. So


Here i'll pick 910pF for the cap since that's a standard value. The R2 value looks to be 22k for this.
Now for R1 you would use the figure 31. Your Fmax/Fmin would be 500k/~2M = 4. 
As previously mentioned, this is an older datasheet and they elected to use log scaling on the graph. I'm interpreting each tick to be an increment of 2 between the 10^0 and 10^1. So your R2/R1 ratio will be ~3.


Given that R2 here is 22kohms and the R2/R1 ratio is 3 -> R1 = 22k/3 ~7.3Kohms. Or 7.5kOhms I believe would be your closest standard value.

So C1 = 910pF, R2 = 22Kohms, and R1 = 7.5Kohms. 
Once again though, if the previous setup worked to their preference, I would atleast first try to reuse that setup with the CD74HC4046A replaced and the old passive components. 

Thanks,
Rami