The DA7 manual is very casual in the section describing connections with Tascam DA-38, -78HR, -88, and -98 decks.
The DA-88 Requires the SY-88 Sync Card
The DA-38, -78HR and DA-98 decks operate perfectly when connected to the DA7 via a simple TDIF 25-pin cable. The DA-88, however, cannot be operated with a simple TDIF cable, because it uses the original version of TDIF, not TDIF-1. TDIF-1 carries a word clock signal on the TDIF cable; the DA-88 expects to find that clock signal on its 15-pin SYNC IN cable. Therefore, the DA-88 must have the optional SY-88 sync card in order to work with the DA7. Ramsa shows the SY-88 installed in its connection drawing, but they don't specifically state that the SY-88 is a requirement.
What Happens If You Try Without The SY-88
My experience in attempting to operate a DA-88 without the SY-88 card was marginal at best. During playback, the symptom appeared to be a lack of Word Clock synchronization, resulting in occasional clicks, pops, and crackles. No amount of Word Clock configuration helped. I tried slaving the DA7 to the TDIF clock, slaving the DA7 to the DA-88's BNC Word Clock output, and slaving the DA-88 to the DA7's BNC Word Clock output. Though these configuration changes resulted in different audio performance, none of them solved the problem.
Attempting to enable digital-in for recording showed that the DA-88 really wasn't going to play along with just a TDIF connection. It was at this point that Ramsa tech support verified the SY-88 requirement. Basically, what the SY-88 is adding is the ability of the DA-88 to slave to external word clock.
Mixed Deck Models
Though I didn't test this configuration, it would seem possible to use a DA-38 or -98 as a master deck, and slave other DA-88s to that master. Proper signals would be arriving at the DA-88s via the SYNC IN cable, so the DA-88s should operate properly with the DA7 via a single TDIF cable. (I have successfully run combo DA-38 and -78HR systems, with either deck as the master.)
Word Length and the DA-38
This may also be applicable to the DA-98; I haven't checked. The DA-38, although it only stores 16 bits on tape, allows you to select 16-, 20-, and 24-bit word lengths to transfer via the TDIF connection. Internally, any word wider than 16 bits is dithered (or truncated, if dither is turned off) down to 16 bits for recording on tape. On playback, the 16-bit word from tape is extended to the chosen TDIF word length. (I presume this is merely extension with right-hand zero bits with no actual improvement in the resolution of the recorded signal. The deck isn't interpolating as a D/A converter would.)
Why Use Longer TDIF Word Lengths?
Within the DA-38, the actual dithering down to 16-bit words happens at the last minute before going to tape. This means that if, for example, you send 24-bit TDIF to the DA-38 during tracking, you will monitor at the full 24-bit resolution---the DA-38 simply echoes back the incoming TDIF signal at full resolution without change. You will therefore get maximum resolution for monitoring during tracking, in case that makes a difference to you.
If you send the full 24-bit words via TDIF to the DA-38, you'd want to enable the deck's internal dithering (unless truncation sounds OK to you). The dithering algorithm on the DA7 isn't described in the manual – I presume it may be a simple random-noise dither like that in the DA-38, though the two may sound different. If you prefer to have the DA7 dither to 16-bit words for transmission to the deck, then you'd set the TDIF word length to 16 bits and turn off the deck's internal dither. Though you'd still monitor at the full TDIF resolution, the TDIF words are now down to 16 bits in this case.
The DA-38 doesn't save the dither state on power-down, but it does save the TDIF width. I caught myself sending the full 24 bits to the deck and the deck had its dither off (the default at power-up). The fact that the deck doesn't remember dither is unfortunate. As a result, I've decided to always have the DA7 dither down to 16 bits, and I set the deck to a TDIF width of 16 bits. Dither is always off at power-up, and now I don't have to worry about Tascam's oversight.
A Warning About the MMC-88
This is a bit off-topic, but there's a bug in the ROM of the Tascam MMC-88 (MMC and MTC/SMPTE box for the DA-88) that prevents it from generating proper MIDI timecode (MTC) at the 44.1KHz sample rate. (The MTC is proper, it just doesn't stay in time with the abs time on the deck.) The MMC-88 works correctly at 48K. I never tested the SMPTE stream to see if that's also affected. This is of little consequence if you're using the DA-88 with the DA7, since you'll need the SY-88 anyway, and the SY-88 can generate correct timecode at both sample rates. The MMC-38, a similar unit made for the DA-38, does not have this problem.